Fullscreen HTML5Med Res (??MB)HTML4

PART 1: PHIL PROSSER Digital Preamplifier and Crossover This advanced preamplifier uses digital processing to provide unprecedented flexibility. It has three digital inputs, including high-fidelity USB, four analog stereo inputs, four stereo outputs, two digital outputs (including USB) and a stereo monitor channel. Individual filters and equalisation can be applied to each pair of outputs, allowing it to act as a digital crossover! Four stereo analog inputs (1V RMS maximum) Frequency response: 7Hz to 43kHz <at> -3dB (with PCM1798 DACs) One analog input can be configured to handle 2V RMS+ S/PDIF coaxial and TOSLINK digital audio inputs Monitor output for analog inputs Four independent stereo output channels, 2V RMS full scale High sampling rate/bit depth USB audio stereo input and output Programmable equalisation, crossovers, relative attenuation & delay for each output Memory for four different configurations Attenuation at 20Hz: 0.3dB; Attenuation at 20kHz: 0.0dB Volume control: +12dB gain to -128dB attenuation in 0.5dB steps Total harmonic distortion plus noise (THD+N): 0.003% across the audio band (largely unchanged to >40dB attenuation) O ur interest in hifi at Silicon Chip runs from simple and ‘purist’ designs such as our Class-A and Class-AB amplifiers, and simple chip-based designs, through to much more complex approaches including high power and even the (very) occasional valve design. We love them all. This author is no exception, owning more audio equipment than most people would consider reasonable, much of it home-built. The larger and more serious hifi setups all incorporate active crossovers, either analog or digital. siliconchip.com.au Our wish-list for an ideal preamplifier includes not only an active crossover but also a USB interface that supports high fidelity playback and recording of music, and of course, switching for four or more analog inputs. This can become cumbersome where the crossover is housed in one box, the USB interface in another, and switching and gain control in another. The aim of this project is to roll all the above into a single ‘Digital Preamplifier’ that fits in a 1U chassis. This allows you to connect your analog audio sources, plug in your USB Australia's electronics magazine connected laptop or phone, TOSLINK source, and provide the functions of a normal preamplifier along with those of an active crossover, equaliser and delay controls for your loudspeakers. For those whose preference is more at the ‘simple is beautiful’ end of the audio spectrum, you may not want to build this device. However, we think you will still find the circuit and other details of this design interesting. That said, if you are into open-baffle speakers, the significant equalisation required for those might make you take a second look at this project. October 2025  29 ADC dynamic range Digital audio provides about 6dB of dynamic range per bit. So the old CD standard of 16 bits gives about 96dB of dynamic range. At the time CDs were released, this was awesome, and even now it is more than sufficient for excellent audio. However, a 24-bit system has more like 134dB between full-scale and the least significant bit (LSB). Consider a real world application like a preamplifier, where the sources can have real impedances, and the ADC sees an input signal-to-noise ratio of, say, 100dB over the 20Hz to 20kHz bandwidth. This is a touch over 16 bits’ worth of digital data above the noise floor. All our remaining bits in the 24bit ADC will be noise plus any signal which may be below the noise floor. If we have the same noise level, but 100kHz of bandwidth presented to the ADC, we will see a noise level about 14dB higher, or 86dB full-scale, in the region of 14 bits’ worth. This is what you see if you look at the ADC I2S data with an oscilloscope. This might sound terrible, but it is not. Remember that the ADC is simply representing the voltage it sees at its input, and these are the peak levels you will see on the SPI data. Most of this is outside the audio band and completely inaudible. The ADC is, in fact, faithfully digitising signals way down in to the -130dB region; way down below the full bandwidth noise floor. There is no question that superb sound quality can be achieved with a decent signal source, a simple preamplifier, basic power amplifier and speaker using a passive crossover. However, the step in capability we achieve in this project through the inclusion of a digital signal processor (DSP) is profound. So come along for the journey of designing and building a no-­ compromise Digital Preamplifier. We will not just present the design, its features, specifications and performance but will also go over some of the challenges faced in pulling together a complex design into something that is reasonably easy to build. The heart of this project is the Analog Devices ADAU1467 IC. This is a 32-bit processor that runs at 294MHz and is optimised for audio DSP tasks. This device has a very rich set of features, including: ● four dedicated stereo inputs and four stereo outputs ● the ability to process data at up to 192 kilosamples per second (192kSa/s) ● the ability to store 400ms of audio data at 192kSa/s ● four stereo asynchronous sampling rate converters (ASRCs) ● an S/PDIF interface ● fully programmable using some very high-level tools We have paired this with: ● a high-quality analog-to-­ digital converter (ADC), the CS5361 or CS5381; the latter provides a better signal-to-noise ratio (SNR) ● up to four high-quality digital-­ to-analog converters (DACs), Analog Devices PCM1798 or PCM1794A chips (the latter provides higher performance) ● a miniDSP MCHStreamer, which provides audio input/output for a computer over USB, making this The Digital Preamp is very capable and compact; its predecessor, which spanned two cases, is shown below. 30 Silicon Chip Australia's electronics magazine preamplifier a very fancy sound card ● switching for four analog inputs ● a PIC microcontroller-based user interface, which allows the whole preamplifier and DSP to be controlled and set up using three buttons and a rotary dial or remote control Everything fits in a single 1U (44mm-high) case, making this a compact and powerful all-in-one preamplifier, switch, crossover and DSP. Simply plug it into your amplifiers and, once set up, all this digital complexity is completely transparent to the user. In my system, this one compact unit replaces the bottom two devices shown in the photo at lower left, saving quite a bit of space! Digital vs analog So, how does this Digital Preamp compare to an analog design? We can hear the wailing and gnashing of teeth from purists at all this digital processing in their signal path. Concern about a DSP like this is ultimately little different to concern about using an op amp in the signal chain. Music & sound has passed through hundreds of op amps by the time it gets to your stereo. In this age, the recording and mixing process includes many DSPs too. No doubt, we need to get the design of a device like this right. But if done properly, the fact that the signal is digitised is transparent. The beauty of it is that, once we have the signal in the digital domain, we can easily apply complex filtering, delay individual channels and implement parametric equalisation with little-to-no reduction in quality. The fact the data is processed as 32-bit numbers means we do not face the old-school challenge of dealing with accumulated errors as we would if we were processing 16-bit data. As an example of the low impact that our Digital Preamplifier has on the audio path, it can take a digital input from the USB port, process it and generate analog output. We can then feed that signal back to an analog input, digitise it and send it back to the PC, all with a distortion result in the region of 0.002%. That’s basically CD quality (which is still considered pretty good these days). This is an interesting test, as it shows that a digital preamplifier can have less impact on the signal chain siliconchip.com.au The Digital Preamp comes with as many inputs/outputs as you would expect from a higher-end system. Each stereo digital output can have different filtering and delay configured, allowing it to also act as an active crossover. than some analog preamplifier circuits. distortion performance. The noise floor is at about -125dB. Performance Overall configuration The performance of the Digital Preamp is essentially defined by the DAC chips used. We measured the performance of the ADC (CS5361/81) using a Stanford Research Labs DS360 signal generator to drive the input to the preamplifier, with the miniDSP monitor output used to analyse the digitised audio. The measured THD was less than 0.0003%, which is consistent with the ‘typical’ specified performance of the ADC chip. Spectral analysis of the ADC data from this test shows no meaningful noise spurs from the switch-mode power supply; 50Hz hum is more than 105dB down. Routing the ADC output to the miniDSP for analysis on a computer, it is clear that the ADC we have chosen is very good indeed, with the distortion products being barely measurable. Turning to the DACs (PCM1794A or PCM1798), their specified THD+N at 44.1kHz is 0.0004%, so slightly higher than the ADC. But we are running them at 192kHz, to allow the preamplifier to operate right across the audible spectrum, and to set aside any concern with bandwidth limitations out to 40kHz. The PCM1794A THD+N operating at this sampling rate is 0.0015%. Our THD measurements are flat at 0.002% across the entire audio band, which is consistent with that. Reducing the volume level with the same input signal by 10dB, 20dB and 40dB shows that the distortion products fall along with the level, so operating the preamplifier with 40dB attenuation has little impact on the The block diagram, Fig.1, shows the signal flows through the Digital Preamplifier. Starting with the inputs, all analog inputs go through a switching section, allowing the chosen one to be digitised using the CS5361/81 ADC chip. The resulting audio data goes into serial port zero of the ADAU1467. Note that the selected buffered analog input is made available on the Monitor output. The ADAU1467 sets the ADC to operate at 192kSa/s. The miniDSP MCHStreamer receives digital data from your PC siliconchip.com.au and delivers data to serial port one of the ADAU1467. The digital audio input can be coaxial (S/PDIF) or optical (TOSLINK). These go to an input switch, which routes the selected signal to an S/PDIF receiver in the ADAU1467. The ADAU1467 chip performs all audio processing, under the control of the PIC microcontroller over an SPI serial interface. Both the miniDSP MCHStreamer and S/PDIF inputs go via their own ASRCs, which synchronise their input sampling rate to the DSP’s sampling rate. Any of the digital, USB or analog inputs can be selected inside the ADAU1467 and routed to the miniDSP Fig.1: the block diagram for the Digital Preamplifier. The digital, USB and analog audio is all routed through the ADAU1467 DSP engine, under the supervision of a microcontroller. Australia's electronics magazine October 2025  31 Soldering the LFCSP-88 ADAU1467 chip We will discuss this more in the construction section (in a later issue of the magazine), but it is worth noting that soldering these ICs is not as hard as we thought. Probably the trickiest part is not putting too much solder on the ground pad. We found we were using far too much, and the IC was floating on it, resulting in poor connections at the edge pads. To address this, after reflowing the IC using a hot air gun, we used a soldering iron to draw a bead of solder along each side, to ensure all the pads were properly soldered. If you use a lot of flux, you can draw a big blob of solder along, and as the pads are small, they don’t have enough surface tension on the solder to form bridges. This quickly solders all the remaining edge pads. You can see the result here. Despite the DSP chip not having any leads, thanks to extended pads on the PCB, it isn’t too difficult to hand-solder. Still, if you are not confident, you’re better off ordering the carrier board with chip already on it. MCHStreamer output. This output goes via another ASRC, which synchronises this output stream to the MCHStreamer sampling rate. The digital audio stream then runs through three parametric equalisers, which operate on the full input data stream, so these affect all output channels. The data is then split into four channels, each being processed similarly. Each has a further three parametric equalisers, followed by crossover filters and delay modules. The four streams are finally routed to the four DACs that provide the analog outputs of the Digital Preamplifier. All DSP processing is done by the ADAU1467 at a 192kSa/s sampling rate, which is just over 5μs per sample. This defines the channel delay resolution and the Nyquist bandwidth limit – though the output DAC analog reconstruction filter has a narrower bandwidth than this. So that filter defines the system’s upper frequency cutoff. Volume control is applied across all channels after all signal processing is complete, and is also implemented digitally. We have measured the performance of volume controls implemented using the PGA2310, a fine volume control chip, and found there to be no real difference compared to using a good 24-bit DAC like the PCM1794/8 and adjusting the volume in the digital domain. While you might worry at reducing the volume resulting in loss of resolution, any spurs and harmonics are so far into the noise floor (below -120dB) that this concern is unfounded. Circuit details Due to the complexity of the overall circuit, and the repetition of certain blocks (specifically the four DACs), we will be presenting the circuit in 10 bitesized chunks. These are spread across two PCBs; eight are on the main PCB, while the other two are the separate main power supply PCB and the front panel controls. The eight circuits that comprise the main PCB are: 1. Analog input switching 2. The ADC 3. Digital audio I/O 4. The DSP core 5. The DACs (four almost identical blocks) 6. The miniDSP interface, which connects to the commercially made MCHStreamer USB interface board 7. The microcontroller section, which includes the LCD interface 8. The onboard power supply, which filters and further regulates the output of the separate PSU board There are a further two circuit sections on separate PCBs: 9. The user controls (rotary encoder, buttons etc) 10. An external AC-to-DC power supply board that feeds the main board We’ll look at each of these in turn. Analog input switching The analog input switching in the digital preamplifier, shown in Fig.2, is pretty conventional. Developing this Digital Preamp required a lot of time and effort; shown in the photo is a prototype that had served its purpose. Yes, I did salvage all the expensive bits... 32 Silicon Chip Australia's electronics magazine siliconchip.com.au All analog inputs have RF suppression beads and 100pF capacitors to reject RF that may be picked up by the input leads or signal source. These have no effect on audio-­ frequency signals. This is followed by 22μF DC-blocking capacitors, biased to ground by a 100kW resistors. These ensure that all inputs have no DC offset, and as you switch between them, there will be no clicks or pops. You will note that on the second auxiliary input (at the top) we have included two optional resistors. If you have a signal source that delivers over 1V RMS, like a CD, DVD or Blu-ray player, you can swap the ferrite beads for (say) 2kW resistors and solder 1kW resistors into these spots. Such a configuration allows for up to 3V RMS without clipping on the ADC. We don’t envisage you will have many really high level inputs, but if you do, you can add similar resistors to other inputs. The -3dB corner frequency of this input stage is defined by the 22μF capacitor and the 100kW bias resistor paralleled with the 47kW resistor at the input to the buffer op amp, which itself has an input impedance of more than 30kW. This frequency is 0.5Hz (1 ÷ [2π {100kW || 47kW || 30kW} × 22μF]), which is way outside the audio band, and will have no impact on audio performance. IC5, an NE5532(A), buffers the input signal and provides a sample of this to the monitor output. It also drives the ADC inputs. You can use a standard NE5532; the A version has slightly better noise limits, although both types have extremely low noise and distortion. We have included 100W series resistors on the monitor output, but remember that this should not be used to drive heavy loads or long lines, as Fig.2: the four stereo analog inputs are routed to the ADC using this circuitry. Switching is via signal relays, followed by an op amp based buffer. siliconchip.com.au Australia's electronics magazine October 2025  33 Fig.3: the left & right channel signals from Fig.2 are digitised here. IC6a/IC6b are inverters that generate complementary signals, which are filtered by IC7/IC8 and clamped by schottky diodes before reaching the ADC chip, IC9. this is a really important signal in your preamp. You will note there are 10W/100μF low-pass filters on the ±10V supply rails. These are included to isolate this section from the other sub-rails that operate from these supplies. The ADC The ADC chip we have selected is the CS5361 or CS5381. The circuit with this is shown in Fig.3. It is pretty much straight from the manufacturer’s datasheet; we have used this configuration in the past with great success. The initial version of the preamplifier used a lower-cost ADC, but we were not happy with the noise floor, so we moved to the tried and true, but more expensive, CS5361/81. 34 Silicon Chip We feel the lower-cost CS5361 is fine for the job, but for a slight premium, you can splash out on the CS5381, which has a 5dB-odd margin in THD+N. Both provide superlative performance. Things to note in this section are the use of the NE5532 dual op amp IC6 configured as a pair of signal inverters. This is required to generate the balanced inputs the ADC requires. We have selected 1kW as the feedback/ input resistance, which finds a good balance between low resistance and thus noise, and ensuring the NE5532 is not loaded too much. Following this are the manufacturer-­ recommended drive circuits, which are unity gain buffers with 91W resistors included to ensure the operational Australia's electronics magazine amplifier is not upset by the notoriously difficult load that the input to the ADC presents. We have spent a lot of time testing alternative ADCs and drive circuits over the years, in the pursuit of low noise and distortion. While the manufacturer’s recommended circuit is fine, we have learned the importance of the 2.7nF NP0/C0G ceramic capacitors across the ADC input pairs. In one test, we tried several different capacitors, ranging from greencaps through silver mica and everything in between. Using a reputable NP0/C0G ceramic capacitor is essential, as distortion increases of over 10dB will be seen if you use something incorrect, such as an MKT capacitor. This is a result of the input presenting siliconchip.com.au a complex load, which will expose even minor non-linearities in these capacitors. Appropriate capacitors are available from the likes of Mouser, DigiKey and element14. You will note that we have another set of local ±10V filtered rails (±10Vfilt2), as we have in all areas of the circuit. This may be over the top, but is a small cost to ensure we have clean rails and minimal risk of noise being coupled between sections of the circuit. CON9 is an I2S test header for the ADC. It is really useful to probe this with an oscilloscope; the LRCLK and MCLK signals in particular. If you are wondering if the ADC is working, trigger your scope off LRCLK and probe SDATA. Note, though, that the ADAU1467 DSP drives MCLK and LRCLK, so do not expect to see anything on these lines until it is up and running. We have included BAT85 clamp diodes on the input to the ADC to protect it from signals that go above the +5V rail or below 0V. This will occur if the input is over-driven, or if an input is connected with a large DC offset. These protect the ADC chip from such excursions. The ADC inputs are internally protected, but we want these as ‘belts and braces’ protection so that your expensive Digital Preamp is safe from abuse. We have tested the distortion performance with and without these protection devices, and there is no measurable difference. The first version of the digital preamplifier used a much cheaper ADC, which we ultimately concluded was a false economy. If you’re going to spend several $100s to build the Digital Preamp, you might as well spend a few more dollars to get the best ADC. of the ADAU1467’s internal ASRCs. These have around 139dB of dynamic range and can up-sample or down-sample with ratios of up to 1:8 and 7.75:1. So we can accept input signals with sampling rates from about 24kSa/s up. When up-sampling, the ASRC generates interpolated data to maintain a 192kHz data stream sampling rate. IC13 is a buffer to drive an S/PDIF output from the digital output signal from the ADAU1467 chip. However, this is not routed to the rear panel, as we don’t have any use for it in our system. It is there if you need it, and it should work (in theory...). DSP core The circuit for the DSP part of the device is shown in Fig.5. The ADAU1467 is an application-specific IC (ASIC) made for audio processing and provides much of the functionality of the Digital Preamplifier. A major reason for selecting this part is that it provides multiple ‘clock domains’, allowing us to integrate the S/PDIF and miniDSP (USB) devices. It also provides sufficient signal processing power and memory for all the volume control, equalisation, filtering and delay functions we require on each stereo band. Once we determined that the ADAU1467 was the right part, we stopped to have a think. This chip only comes in an 88-lead, 12 × 12mm LFCSP package with a 5.3mm square exposed pad underneath. We put in a lot of effort to stick to through-hole parts where we can, and Digital audio I/O We have included S/PDIF (coaxial) and TOSLINK (optical) digital audio receivers, and included the ability to decide which goes to the ADAU1467 DSP. It includes a receiver that can handle the raw (low-level) signals from a coaxial link. The switching circuitry is shown in Fig.4. The clock for the digital audio stream is generated by the signal source. This means that we need to synchronise the input clock source to the Digital Preamplifier clock source; otherwise, we will end up with more samples than we need, or not enough. For this we use another siliconchip.com.au Fig.4: the digital I/O is quite simple as there are just two digital inputs (one TOSLINK [OPT1], one S/PDIF [CON10]) that are selected by a single relay. The outgoing digital signal is fed directly to OPT2, and to the S/PDIF output RCA connector via buffer IC13 and a 75W impedance-matching resistor. Australia's electronics magazine October 2025  35 when forced to use SMD parts, endeavour to use manageable packages, selecting the largest lead pitch we can. This part not only has a ‘fine lead pitch’, it doesn’t even have leads! The project kind of sat on the shelf for a while, and in the end Phil decided to build the Digital Preamplifier for himself, as a lot of the design and software was ready to go from previous designs. He toyed with the idea of going back to one of the older ADAU devices that he has used in the past, but this would have demanded compromise on performance, and we really wanted to use a recent device. 36 Silicon Chip As it turns out, soldering the chip was not as hard as he initially thought (see the accompanying panel on page 32). We also found a way to avoid soldering it entirely if you are dead set on that! The layout and peripheral components around the ADAU1467 (IC18) are straight from the data sheet. Besides the support components, mostly this part of the circuit is just routing signals to and from all the other parts. All the components around IC18 are surface-­mounting types, because the chip runs at a high clock rate (nearly 300MHz) and needs excellent Australia's electronics magazine local filtering of the supply rails. We have stuck to M2012/0805 devices where we can; they are massive (2.0 × 1.2mm) compared to the lead pitch on the chip, anyway. We have included a clock buffer for the system master clock (IC10), which runs at 24.576MHz. This distributes this clock signal to the ADC and DACs. We have also included a header for probing the SPI interface between the PIC microcontroller and the ADAU1467. This is mainly for debugging, but you might find other uses for it. When we were building the first prototype of the Digital Preamplifier, we siliconchip.com.au The Core boards cost about $80 at the time of writing, which is a bit of a premium on the $30 cost of parts from Mouser/DigiKey. Still, if you do not feel confident in soldering the chip, we recommend you shell out for one of these. We tested two, and both worked fine. DACs Fig.5: the DSP core is where all the digital signal processing occurs. It’s little more than IC18 and its support components. If you don’t fancy soldering IC18, you can buy it on a carrier board and plug it into the two headers shown at upper-right. In that case, none of the other parts shown here but IC10 are installed. became aware of the “ADAU1467 Core Board” and development boards on eBay and AliExpress. This board is pretty much exactly the same as our ADAU1467 core circuit, which we replicated from the OEM design notes. So much so that we were able to buy one, and ‘graft’ it onto our board, simply leaving off IC18 and its support components. So we rolled this into our design, and now you are able to choose whether you solder that 88-pin leadless chip, or leave the whole section off and plug in a purchased ADAU1467 Core Board to the two DIL headers shown at upper-right in Fig.5 (and in this photo). siliconchip.com.au There are up to four onboard DACs, all based on PCM1794A or PCM1798 chips. The circuit for one of these is shown in Fig.6. These chips are pin-compatible, with the PCM1794A being more ‘premium’, offering 127dB dynamic range vs 123dB and a THD+N of 0.0004% vs 0.0005% at 44.1kSa/s. We are running the whole digital signal processing part of this design at 192kSa/s for a couple of reasons. Firstly, if we want to implement time alignment with simple buffer delays applied to individual channels, the delay resolution is defined by the DSP clock rate. 192kSa/s is 5.2μs per sample. It would be possible to implement a filter to generate this phase shift at a lower sampling rate, but that would substantially complicate the programming and affect phase linearity across the band. A second reason for using a 192kSa/s sampling rate is to ensure that the frequency response is flat for the entirety of the audio band and well beyond. We want the Digital Preamplifier to be as transparent as practical. We have used the CS4398 in several other designs with great success. However, while developing this Preamp, stocks were low and lead times long. So we went to the PCM1794A/98. If you look at the datasheet for this DAC, you see excellent specs, a dynamic range and signal-to-noise ratio of 123dB (129dB for the PCM1794A), and a THD+N of 0.0015% for both. Hold on, didn’t we just say that the THD+N was 0.0004%? Looking more closely at the datasheet shows that this is true at 44.1kHz but at 88kHz, the THD+N doubles to 0.0008%, and at 192kHz, it nearly doubles again, to 0.0015%. None of these are even remotely a problem, and the dynamic range of these chips is even better than our usual CS4398. With their superlative SNR, they are very well suited to our application, where we will be performing volume control digitally. We spent some time measuring this sampling rate dependency of the THD figure. Especially given that super low noise floor. With a 1kHz, 1.8V RMS output (0.9V RMS input), the second harmonic is at 0.0015%, which is entirely consistent with the specified performance. The noise floor is about 130dB below full scale. The Digital Preamplifier design keeps wiring and cabling in a high-end hifi system to a minimum. This version uses the plug-in ADAU1467 Core Board rather than a discrete chip. Australia's electronics magazine October 2025  37 We found some low-frequency spurs that were mains-related and might be because the Digital Preamp sat on top of the signal generator during testing. As with the ADC, we have included an I2S header (CON1) on each channel. These provide test points for the MCLK, LRCLK, BCLK and SDATA signals. Once the DSP chip is running, you should see a 192kHz square wave on the LRCLK line, with the data and other clock signals synchronised to it. The circuit is pretty much what’s recommended by the manufacturer, and as it does what it says on the box, we see no need to change it. A relay is included for each channel that disconnects the output at power-up and power-down. This prevents unwanted noises being sent to the speakers. USB interface This is a somewhat expensive, but we think really important, component of the Digital Preamplifier. It allows high-quality audio to be received from and sent to a computer via a USB port. It does this by interfacing to an external board. The interface is electrically isolated, as shown in Fig.7. The MCHStreamer Lite (which excludes the unnecessary optical input) costs ~$150. We have seen several other USB-to-I2S data converters, but no alternatives at a good price that can also perform the I2S-to-USB task. If you do not need to record audio from your Digital Preamplifier on your Fig.6: one DAC channel; the op amps and associated resistors and capacitors form the reconstruction filter. This circuit is replicated four times on the board, with only the DAC_SCLK_CHx, DAC_DATA_CHx and physical output connector varying between them. 38 Silicon Chip Australia's electronics magazine siliconchip.com.au computer, you could substitute the miniDSP MCHStreamer with an ‘output only’ alternative and wire it into the miniDSP headers. The MCHStreamer interface has been kept as simple as possible. The headers allow the MCHStreamer to be connected to the Digital Preamplifier board using flying leads. Only a handful of wires are actually needed, but to keep things tidy, we used the plugs and flying leads miniDSP provided and soldered all the pigtails to the PCB. We have placed the connectors on the board so that if you solder the pigtails with the connector standing straight up from the PCB, the pigtails simply go straight down into corresponding PCB pads without any wires crossing over etc. You then bend the wires to plug into the miniDSP as shown in the photos. The MCHStreamer deals with the USB to I2S conversion. On a Windows computer, you need to install ASIO drivers; once you have made the purchase from miniDSP, they are available for you to download and use on all Fig.7: this circuit snippet interfaces the MCHStreamer USB audio I/O interface with the rest of the circuitry. It’s isolated to avoid hum loops and such. your devices. On Linux and Mac computers, the device will simply work. The MAX22345SAAP+ isolates the computer’s USB port from the Digital Preamplifier. This avoids annoying hum loops, which are a notoriously common with laptops and PCs. This isolation is for noise reduction only; it is not galvanic isolation to provide mains or high-voltage protection. Microcontroller The microcontroller circuit shown in Fig.8 does a few things: Fig.8: the microcontroller circuit, which configures the DSP and handles the user interface. The buttons and rotary encoder connect via CON16, while the alphanumeric LCD is wired up via either CON8 or CON19. siliconchip.com.au Australia's electronics magazine October 2025  39 Replacing the PCM1798 with a PCM1794A While the Preamp can be built with either PCM1794A or PCM1798 DAC chips, and they are pin-compatible, some components need to be changed; the circuit is shown with values to suit the PCM1798. The reason for this is that the full-scale output current is different, being 7.8mA for the PCM1794A ($25 per chip) and 4mA for the PCM1798 ($10 per chip). To change from the PCM1798 to PCM1974A, there are a handful of resistors and capacitors that need to be different values, and a couple of parts that are omitted. These are listed in the notes in Fig.6 and on the PCB silkscreen. ● It loads the required software into the ADAU1467 on power-up ● It displays information on a 16×2 character alphanumeric LCD ● It handles sensing for the pushbuttons and rotary encoder ● It initialises and communicates with all the other chips, like DACs and ADCs ● It handles input selection, volume control, equalisation etc ● It decodes and handles infrared remote control signals The chip (IC15) comes in a 44-pin QFP package that is not difficult to solder. We thought about implementing a fancy graphical display, but there’s pretty limited space on the front of a 1U case, which is just 44.5mm tall. While the Digital Preamp is a fairly advanced piece of gear, the alphanumeric LCD provides enough space to do what we need, ie, adjust volumes, switch between inputs and set up digital filters. The user interface to the Digital Preamplifier needs to: ● Let you set up the channels in terms of crossover parameters, slopes, relative attenuation and frequencies ● Let you set up the equalisation ● Let you set the subwoofer channels for mono or stereo output ● Select the channel to monitor ● Select the input to listen to ● Change the volume Once the unit is set up, it is only the last two things you will ever really do. We generate a negative bias voltage for the LCD from the -10V rail using a simple LED voltage drop (LED2). We need this as we are running the 16×2 LCD from just 3.3V, and the panel needs close to 5V on the bias to operate properly. Every 16×2 LCD we have seen works well this way, and this makes the LCD data interface compatible with the 3.3V PIC microcontroller. 40 Silicon Chip A typical 16×2 LCD screen has a 16-pin SIL interface, which we have adapted for convenience to an 8×2pin header (CON19). This way, we can crimp an IDC plug onto a ribbon cable and simply plug it into the PCB. The wires at the other end can then be soldered to the LCD’s SIL header, or via another IDC plug and a small adaptor board that we’ve used before. Controls The controls (buttons, rotary encoder etc) are mounted on a small, separate PCB; its circuit is shown in Fig.9. The board houses three push button switches, a rotary encoder with an integrated pushbutton switch, and a TSOP4136 infrared (IR) receiver. This mounts to the front panel using the rotary encoder boss and nut. The rotary encoder on the front panel is a volume control most of the time. There are two buttons to the left that let you switch through the available inputs. The GUI defaults to showing the volume and input selected. If you push the volume control in, it will save the current parameters. If you push the button to the right of the control, which is like a ‘back’ button, you can rotate through the other menus, which allow you to change: ● Crossover parameters ● Equalisation ● Load a setup ● Save a setup to one of three spots On power-up, the system reads the configuration from its EEPROM. There will not be valid data on the first power-­­up, so the software will use default values. Remember to save your setup once you enter it; after that, the system boot to your main configuration on power-up. Power supply All that remains of the circuit is the power supply. This is split into two parts, because the main rectification, filtering and pre-regulators are on a standalone power supply board. We have done this to ensure that all the rectification and switching ‘stuff’ happens away from the mixed signal analog and DSP board. It also means that if we want to change the packaging or power supply, we can do this simply. The circuit of this separate power supply board is shown in Fig.10. This is pretty conventional; it generates a 5V digital supply and ± 10V DC rails for the analog parts of the Digital Preamplifier. The main challenge here is the need for well over 250mA from the analog rails and in excess of half an amp on the 5V rail. This makes it difficult to design it to run from a DC supply, with a voltage inverter generating the negative rail. It makes using a single AC input (such as from a plugpack) less than a great idea. We were using a 16V AC 1.38A plug pack in this way during tests, and when the DSP was loaded, the plugpack fuse blew! The plugpack Fig.9: the small control board circuit; CON1 connects directly to CON16 shown in Fig.8. Australia's electronics magazine siliconchip.com.au also produced high (±22V) unfiltered analog rails, resulting in high dissipation in the regulators. Instead, we are using a dual 12V AC secondary 30VA mains transformer to drive the power supply board, mounted in the same case, near the supply board. REG1 & REG2 need heatsinks; they will get toasty warm, but they do pass our ‘can you hold your finger on them’ test. The digital rail uses a switch-mode buck (step-down) converter. This is required to efficiently drop the unregulated 16V rail down to a regulated 5V. The LM2575-5 (REG3) does not get hot and can operate without a heatsink. We have used generous main filter capacitor banks, with three 2200μF capacitors per side. You could probably get away with half that; the main reason for using this many was the ripple current. Two 1000μF capacitors were within specification in this circuit based on their ripple current rating, but they got warm during operation, which does not bode well for a long service life. So we switched from two to three devices and (more than) doubled their capacitances to be safe. The regulators are fed through 47μH/100μF LC low-pass filters. These, along with cuts in the ground plane, seek to isolate digital current paths to the main filters from the analog regulators. Onboard regulation The +5V, +10V & -10V supplies from CON2 & CON3 on the power supply board are fed to CON12 & CON11, respectively, on the Digital Preamplifier board – see Fig.11. The main digital supply is +3.3Vdig. This is generated from the incoming 5V rail using a low-drop out regulator (REG1, LD1117V33). This is distributed on a power plane on the fourlayer PCB (more on that later). The ADAU1467 DSP also has an analog 3.3V input, which we don’t really rely on, but we have included a separate regulator to provide clean power to it (REG2). We figured if we left this off, we would regret it at some point! It just depends on how the software in the ADAU1467 is written. The ±10Vfilt1 rails are simply filtered versions of the ±10V supplies from the power supply board. As we’ve seen in the other circuits we’ve looked at, many of them have additional filtering to feed the individual ICs. The 5V analog rail for the DACs (+5Vdac) is derived from the +10Vfilt1 rail, as we want this to be as clean as possible, and definitely do not want digital or switching noise from the other 5V rail creeping in. The power supply also includes circuitry to control the output-enable relays (the bottom third of Fig.11). This holds the output relays off during power-up and disconnects the outputs as soon as power is removed. This is extremely important when driving a power amplifier directly, as we need to suppress any start-up and shutdown ‘thumps’. There are two main sources of these; the first is the operational amplifiers and DC decoupling capacitors settling. The second is the ADC and DAC, which use a single-rail analog stage with the input and outputs offset by 2.5V. This offset is removed by Fig.10: this separate power supply board converts the 2 × 12V AC inputs from a toroidal transformer to the +5V and ±10V rails that power the whole Digital Preamplifier. Note that the case is Earthed and the PCB Earth connection is via one of the PCB’s mounting holes. siliconchip.com.au Australia's electronics magazine October 2025  41 AC-coupling the signals, but charging and discharging these capacitors takes a little time. The start-up circuit monitors the ‘half rail’ voltage between the positive and negative rails, via two 4.7kW resistors near the centre of the circuit. This is compared to the same half-rail voltage but filtered by a 220μF capacitor. Q2 and Q11 together sense a difference in excess of ±0.6V, and if this is detected, they switch on Q13, which disables the output. This capacitor is discharged at powerup, so it ensures the system is muted then. Also, as the rail voltages drop at power-off, this holds charge and forces the output to be muted as soon as one of the rails has dropped by 0.6V. PCB design As briefly mentioned earlier, the rather large main PCB is a four-layer design (the power supply needs only two layers). The main advantage of doing it this way is that we can have two signal layers (on the top and bottom of the PCB) and power/ground planes on the internal layers. This greatly simplifies the job of routing the PCB, as we need to do very little to correctly connect the power and ground pins of most components. It also keeps voltage drops nice and low. Fig.12 shows the power plane with multiple rails. These allow distribution of the digital and various analog rails to each section of the circuit. As we’ve seen in the circuit diagrams, each main analog section has its own sub-rails derived from the ±10V rails using 10W/100μF low-pass filters. The blue areas in Fig.12 show the internal layer that distributes the various power rails. The pink area is the ground plane; it extends throughout the whole of the blue area, too. The top plane of the PCB is primarily digital traces (orange/brown), while the bottom plane of the PCB primarily carries the analog signals (mauve). Once we get power and ground traces onto their own planes, routing becomes a lot easier, and we are able to choose optimal routing of signals without the need to accommodate those power and ground traces at the same time. Fig.13 shows the copper traces with the power plains hidden. Here, you can see how we have separated the digital and analog sections of the circuit. You will also note the differential output lines near the DAC chips (towards the upper right) running close together in pairs. This has been done throughout the layout to minimise hum and noise pickup. This extends to the output. Similar attention has been paid to the input stage and ADC. Packaging We have used an Altronics H5031 one-rack-unit (1U) case to hold all Fig.11: the Digital Preamp’s onboard power supply circuitry. This includes a filter for the ±10V rails, a +5Vdac analog supply that’s derived from them, two +5V and two +3.3V digital rails derived from the incoming 5V supply, plus the power on/off output disconnection control circuitry shown at the bottom. 42 Silicon Chip Australia's electronics magazine siliconchip.com.au this. It is very neat and not hard to do the metalwork for – although there is a fair bit of drilling on the rear panel. The rear panel houses the IEC mains connector, mains fuse, holes for the USB & S/PDIF inputs, plus 10 dual RCA connectors for the analog inputs and outputs. Next month That’s all we have space for in this issue – that was a lot to take in at once! Next month, we will present the parts list, PCB assembly and initial testing instructions. After that, the third and final part will cover case drilling & cutting, final assembly, wiring and usage SC of the Digital Preamplifier. Figs.12 & 13: by making the PCB a four-layer design, we have the luxury of an internal power and ground plane, along with the top/bottom layers, which are used mainly for signal routing. The blue areas are the internal copper pours for power distribution, while the pink area is the internal ground plane (left diagram). The diagram on the right shows the board without the power planes, so you can see the top & bottom layers more clearly. You can see how clean the signal routing is, since power and ground tracks are not needed on these layers. We keep the digital and analog tracks separate by routing them on opposite sides of the PCB. Both diagrams are shown at 59% of actual size. siliconchip.com.au Australia's electronics magazine October 2025  43