Fullscreen HTML5Med Res (??MB)HTML4

ASK SILICON CHIP Got a technical problem? Can’t understand a piece of jargon or some technical principle? Drop us a line and we’ll answer your question. Send your email to silicon<at>siliconchip.com.au Battery Balancer Mosfet inverter/driver query I would like to raise a potential problem with the use of the N- & P-channel Mosfets in the QS6M4 device as a complementary pair in the design of the High-Current Four Battery/Cell Balancer (March & April 2021; siliconchip.com.au/ Series/358). They are connected back-to-back across the 3.3V supply, so the only resistance in the circuit is the devices themselves. The N-channel gate threshold is 0.5-1.5V, while the P-channel is -0.7 to -2V. So potentially, from 0.5V to 2.6V (3.3V − 0.7V), both transistors are on. At -1.5V, the P-channel RDS(on) is 0.155W, and at 1.5V, the N-channel RDS(on) is 0.17W. So, if you simplistically apply V = IR, you get I = 10.15A = 3.3V ÷ (0.155W + 0.17W). The maximum pulsed drain current for both devices is 6A. You limited the surge current through Q11a/b into Q10 with a 1W resistor. I’m wondering if the complimentary pair should have similar protection, even though the ‘both on’ time is short. Could it shorten the life of the QS6M4 otherwise? (D. H., St. Ives, NSW) • Duraid responds: I spent considerable time validating the use of these Mosfet pairs during the design phase. Shoot-through currents in a CMOS inverter are a concern, especially at higher temperatures where the gatesource on-thresholds start creeping down. The 0.155W/0.17W resistance values quoted are for devices that are well and truly on. Resistances around the 1.65 (Vdd ÷ 2) point are nearly an order of magnitude higher, especially for the PMOS section. Keep in mind that the gate threshold voltages quoted are typically for channel current flows of around 1mA, not the many amps that the devices are capable of with higher gate-source voltages. At these low gate voltages, it’s thersiliconchip.com.au mal degradation that you need to be careful to avoid. These parts claim to be able to tolerate 1W, so long as thermal limits are observed. During the design phase, I ran some simulations which showed an order-of-magnitude headroom to this limit, and significant thermal headroom, even allowing for the fact that these parts are quite near the inductors/power FETs. So to summarise, you do have to be careful using Mosfet pairs as inverters like this, but we have verified that these particular parts are suitable in this configuration. Sourcing LM5163 for Battery Multi Logger The LM5163DDAR buck converter (regulator) IC used in the Battery Multi-Logger seems to be in short supply. Digi-Key and Mouser are both quoting a seven month lead time. Do you know of an alternate supplier or device? (R. M., Paynesville, Vic) • You can use the automotive version, which is identical to the part we specified, just a bit more expensive: LM5163QDDARQ1 or LM5163HQDDARQ1. It is in stock at both Digi-Key and Mouser. Reed relays are underrated for 1A PSU I am reading my way through the February 2021 issue of Silicon Chip. I have a couple of comments about the article on the Arduino-based Adjustable Power Supply, starting on page 38 (siliconchip.com.au/Series/357). There is no back-EMF diode across the coil of the reed relay. Surely one is required to prevent damage to the Arduino pin. Also, both the Altronics and the Jaycar reed relays have a contact rating of 0.5A. So rating the power supply at 1A and expecting the relay to make and break that current is not a good idea. My experience is that reed relays will tend to stick closed if overloaded. I note Jaycar do sell a reed relay with Australia’s electronics magazine 1A rated contacts, Cat SY4036. Altronics don’t seem to have any 1A-rated reed relays. Finally, I am looking forward to reading the article on the computer upgrade. It is good to have this as I am interested in learning of the potential ‘snags’ when doing so. (D. W., Hornsby, NSW) • Regarding the relay coil, since the Arduino pin is pulled to ground and effectively shorts the coil terminals when switching it off, there is no opportunity for a high-voltage spike as would occur if the circuit were simply opened. The circulating current decays via the driving pin, so it never goes above the coil operating current. We have never seen this sort of arrangement fail. It’s when you have an open-collector or open-drain relay coil driving arrangement that the back-EMF quenching diode is needed. You are probably right about the Jaycar/Altronics relays being slightly underrated for this project. We have used similar relays in the past rated to break 1A, but as you point out, the ones we specified are only rated to carry 1A. Fortunately, these relays have a quite high (100V) voltage limit, and the current limit can be set in the PSU to provide an extra degree of protection. Oddly, the Altronics Cat S4100 & Cat S4101A relays have a rated switching current of 1A but are described as “0.5A 5VDC SPST DIP PCB Mount Reed Relay”. We aren’t sure if that is a mistake, or if they can actually break a higher current than they are rated to carry continuously. Panel meters fail when used with inverters After reading Jim Rowe’s review of mains panel meters (December 2020; siliconchip.com.au/Article/14678), I bought a PZEM-051 panel meter to fit to a portable power supply. This consists of two deep-cycle AGM batteries wired in series with provision to connect up to three 24V DC to 230V AC inverters. Two of the inverters are April 2021  107 modified square wave types, while one is a pure sinewave type. The panel meter immediately failed. After checking the wiring carefully, I removed the back and checked for signs of faulty components. Q1 (a 600V Mosfet) looked stressed. Tiny globules of solder were stuck to the Mosfet and the adjacent PCB. When power was again applied, the Mosfet became too hot to touch within seconds. I contacted the supplier, who eventually advised that the meter was not designed for use with inverters, although there was no information about this in the supplied instructions. Is it possible to fit a transient voltage suppressor across the 24V power supply to a new meter to prevent another failure? The Jaycar Cat ZR1152 TVS looks good. I might have to fit two in series to prevent premature tripping, though. Any suggestions you can provide will be much appreciated. (I. M. P., Fullarton, SA) • The waveform from an inverter, especially a modified square wave type, has much greater harmonic content than a normal mains waveform, so we are not surprised that this could damage a low-cost panel meter. Adding a TVS across the DC side of the inverter probably won’t help, as it is likely the spikes and steps in the ‘230V AC’ waveform that are causing the damage. You would need to filter that waveform before feeding it to the panel meter(s). However, finding a filter that will remove enough harmonic content to keep the panel meters safe, without damaging the filter itself, might not be easy. We suggest that you try using this Jaycar EMI filter (Cat MS4001) between each inverter’s output and the panel meters/outputs. How much to build the USB SuperCodec Can you give me an approximate costing for parts to build the USB SuperCodec (August-October 2020; siliconchip.com.au/Series/349)? (R. P., Tea Gardens, NSW) • Phil Prosser added up what he paid for all the parts to build the prototype, and came up with a figure of $439.28, including the power supply and case, but not including the PCB, for which we charge $12.50 plus delivery costs. Not bad, we think, considering the resulting performance. 108 Silicon Chip High Power Ultrasonic Cleaner not working I have built the High Power Ultrasonic Cleaner (September & October 2020; siliconchip.com.au/Series/350), but I am having trouble getting it to work correctly. After setting it up, I switched on the unit and checked the 5V volt supply at IC1 and IC2. I got a reading of 5.04V. I then filled the bath with 3.5L of water, switched it on and tried to calibrate it. The 25% and 50% LEDs gave a brief flash, then the run LED lit up, but the unit would only run at 10% power. I checked the connections to the transducer; they all appear OK. I initiated the diagnostic mode and could only get a maximum reading of 2V at TP1. I rewound the transformer, adding an extra layer of 28 turns. This time, when calibrating, the 25% LED stays on, and the 50% LED pulses every two seconds. After approximately two minutes, the run LED lights up, but it will only operate at 25% power. In diagnostic mode, I now get a maximum reading of 4.8V, at which time the unit goes into current overload. I tried altering the quantity of water in the bath, to no effect. I removed ten turns/windings from the transformer. This dropped the maximum voltage reading on TP1 to 3.7V, but made no change to the calibration or running of the unit. Do you have any ideas? (P. H., Mosgiel, New Zealand) • It sounds like the transducer resonance point is not being found. Try running the diagnostics and sweeping the frequencies manually to find the maximum current by measuring the voltage at TP1. If this voltage goes over the 4.8V maximum, reduce the number of secondary turns on the transformer. The number of turns needs to be such that the current limit isn’t reached at resonance. This is the only way to find the transducer resonance frequency correctly. Then the cleaner should then run correctly, and you can achieve the ultimate power by altering the transformer secondary windings, which should be within a few turns of the ideal number once you are reaching resonance without overloading it. Tapped transformers with 45V Bench Supply I have just ordered the parts and PCB Australia’s electronics magazine to build your 45V 8A Linear Bench Supply (October-December 2019; siliconchip.com.au/Series/339). The circuit design looks good to me, but I’d like to make a few modifications to reduce heat dissipation for my use. I built a number of the older ETI-163 supplies many years ago. That design used multiple winding on the transformer, switching them in series as the rotary potentiometer was rotated on the front panel. Is there a reason why you didn’t use a similar approach for your supply? I designed my own version using three separate 14V 10A transformers switched to series or parallel combinations. I chose 14V as those combinations are a few volts above the most commonly used voltages for my industry, 13.8V and 28.8V. I used a simple op-amp voltage divider to switch the windings based on the ‘selected’ voltage on the front-mounted voltage pot. My voltage/current regulation was based on the old ETI-163 power supply. At 13.8V DC output, the transistors were only dropping 5.9V, so at higher currents, the heat dissipated was minimal (60W at 10A or 120W at 20A). This also has the advantage of lowering the output impedance of the ‘source’ as there are two windings in parallel (great for high current loads). At 28.8V DC output, two of the transformers are connected in series, with one unused. Past 32V, all three windings are in series, giving up to 60V <at> 10A before the series pass transistors. Using eight MJE15003 transistors on two large heatsinks with 2 x 80mm fans, the heat was spread out quite well, and I have never encountered any overheating problems. However, at 16V DC, the heat output is quite significant at 235W with a load drawing 10A. I also had a 0.5A/10A range select switch, which switches a different shunt in the negative line to allow fine current limit control at lower currents. Metering was analog like the ETI-163 as they are fast and easy to read at-aglance, especially the current meter. I’ll probably add this feature to the new supply, maybe with three current ranges: 0-500mA, 0-1A and 0-10A. (B. N., Marine Terrace, WA) • We did consider using a multi-tap/ series/parallel transformer configuration while designing the 45V PSU, but we couldn’t find any suitable offsiliconchip.com.au the-shelf transformers at reasonable prices. We didn’t want to use multiple transformers as that would result in a much bigger, heavier unit. The switchable shunt idea is interesting, although you’d have to have your wits about you to know what range you were using at any given time. Also, the switch resistance could introduce some inaccuracies, and possibly unreliability long-term. have not tested it with Python 3. The error “ImportError: No module named ‘urllib2’” confirms this, as per the following StackOverflow question: siliconchip.com.au/link/ab76 If you want to push ahead and try to make it work with Python 3, the advice on that web page is a good start. Tide Chart Python version mismatch A colleague (who does not read your magazine; shame on him!) has an appliance where the backlight has failed on the LCD panel. Am I correct in assuming that the light is integral to the panel, and hence cannot be replaced? I suppose that by squinting at the panel, or perhaps by shining a bright light upon it, the segments could be discerned. (D. H., North Gosford, NSW) • It depends on the LCD panel. You I am trying to get the code for your Raspberry Pi Tide Chart (July 2018; siliconchip.com.au/Article/11142) running, but I am getting an error “ImportError: No module named ‘urllib2’”. (P. C., Balgal Beach, Qld) • We suspect that you are trying to make the Tide Chart work with Python 3. It was written for Python 2, and we How to fix failed LCD backlight can experiment by applying light to the panel using a small torch. There might be a way to provide backlighting by feeding light in from the panel’s side or back. Front-lit LCD panels are harder to control for lighting, but you may get sufficient display brightness with front lighting. The display contrast is usually poor with front lighting. We occasionally publish entries in Serviceman’s Log where contributors have successfully replaced the backlighting on various LCD screens. You really have to open it up to see whether it is possible for that particular display (unless you can find information about that aspect of it online). Ferrite bead selection for amplifier The Ultra-LD Mk.4 200W RMS Power Amplifier (August-October How is negative feedback affected by phase shift? When feedback is being discussed, the effect of phase shift on a feedback loop is usually considered, but always in the most extreme situation where the phase shift is large enough to set up positive feedback and drive the circuit into oscillation. But surely, any phase shift should have a detrimental effect on feedback since phase shift is caused by a time delay in the feedback circuit. That in turn means that the circuit is feeding back an error to a different part of the signal; in effect, trying to correct an error that has already happened and the source signal has moved on. And yet, the vanishingly low distortions being measured in some high-end amplifier circuits, like those published in Silicon Chip, suggest that this is not happening. Can someone explain why feeding back a delayed signal is not a problem for a feedback circuit? (P. T., Casula, NSW) • Yes, phase shift has a detrimental effect on negative feedback used for distortion reduction or accurate gain setting. It’s worse at higher frequencies as the circuit will typically have a fixed feedback delay, representing a larger phase shift relative to higher frequency signals. This is largely siliconchip.com.au why audio amplifiers usually have rising distortion with frequency, typically evident above 1kHz. Therefore, audio amplifiers usually are designed to operate just on the edge of stability, with the minimum possible delay, pushing this point of rising distortion above 20kHz where it is not audible (and the amplifier will generally be designed not to reproduce signals above this frequency). Consider that the open-loop bandwidth of an audio amplifier will typically be in the megahertz, yet it is only tasked at reproducing frequencies (in closed-loop mode) up to 20kHz. So if the phase shift is, say, 90° at 2MHz, that equates to a feedback delay of 125ns (90° ÷ 2MHz ÷ 360°). For a 20kHz signal, that’s a phase shift of 0.9° (360° × 125ns × 20kHz). Therefore, the negative feedback is still more than 99% effective, reducing the open-loop distortion by more than 40dB. As long as the design is fairly linear (ie, open-loop distortion is not gross), this is usually enough to give a very low distortion figure even at 20kHz. If you look at the evolution of our amplifiers, 20 years ago, we were achieving figures of <0.001% <at> Australia’s electronics magazine 1kHz, but significantly higher (say, between 0.01% and 0.1%) at 20kHz. These days, the open-loop bandwidth has been raised, making feedback more effective; open-loop linearity is better, and other factors have been improved to the point that we are achieving close to 0.0001% <at> 1kHz and still well under 0.001% <at> 20kHz, leaving little room for further improvement. So you are right, the phase shift in a negative feedback circuit is undesirable, but luckily, it can be kept to a low level where it is not bothersome. In circuits like low-pass and highpass filters that inherently have a phase shift within the audio frequency band, the linearity of the change in phase with frequency is usually excellent. We make sure that it is by using all linear components in the RC networks, and so it does not introduce harmonic distortion. It does introduce a frequencydependent phase shift, but in theory, for normal ‘listening’ conditions, this is inaudible. It can cause problems in certain scenarios like interactions between drivers in multi-drive loudspeaker systems, in which case, the crossover circuit design can be critical in achieving good results. April 2021  109 2015; siliconchip.com.au/Series/289) uses an SMD ferrite bead. What value of inductance/resistance should it have? There are many to choose from at Digi-Key. (I. G., Oak Flats, NSW) • The ferrite bead type is not critical. Ferrite beads don’t have any significant inductance or resistance. They are usually specified with an impedance in ohms at 100MHz. The cheapest from Digi-Key in the M3216/1206 package are rated at either 600W or 1kW at 100MHz, and either would be fine. For example, the Bourns MH3261-601Y or Eaton MFBM1V3216-102-R. Component damage in CLASSiC DAC? I have finished building your CLASSiC DAC from the February-May 2013 issues (siliconchip.com.au/Series/63). I went through the testing procedure, and everything was fine in regards to the power supply until I bridged LK1 and LK2. The DAC chip heated up rapidly, so I inspected the board and found I had accidentally fitted TOSLINK transmitters and not receivers. I have since replaced them with the correct receivers and carried out the setup procedure again. But I still have the same problem with the CS4398 DAC chip rapidly heating. Upon further investigation, I found that when the JP1 link is set for 3.3V, the unit will continuously scan the four sampling rate LEDs and not detect any channels, and there is no audio output. The DAC chip does not get hot. When JP1 is connected to 5V, the unit does select channels and detects the sampling rate, but that is when the DAC chip gets very hot in a matter of seconds. There is audio output, but it is very noisy. I have also tried with JP1 out and the unit powers on fine, selects all channels and audio from USB and SD card can be played back, but there is a lot of noise through both the headphone and line outputs. However, the noise is not as bad as when 5V is selected. Could it be that having the wrong TOSLINK transmitters/receiver fitted has damaged the CS8416 receiver chip and introducing the noise to rest of the circuit? I would love to hear your thoughts before I purchase a new chip. (J. R., Warrane, Tas) • We can’t see an obvious way that fit110 Silicon Chip ting TOSLINK transmitters instead of receivers would damage anything. We wonder if you have another problem and the TOSLINK transmitter error is just a coincidence. Do make sure that the TOSLINK receivers you have fitted are the right type, though. The only thing that jumper JP1 controls is the voltage fed to Q13b, which then goes to the TOSLINK receivers and nowhere else. Their outputs are AC-coupled to the CS8416, so it should not be possible for the wrong voltage to be fed back. We suspect you have a short circuit from some point on this rail to something that feeds to the DAC, such as the +3.3V rail. This short could be between the pins of JP1, or perhaps between some pins of Q13. It could be elsewhere, but we can’t see any other obvious locations. We suggest removing the jumper from JP1 and check for continuity between the middle pin and both of the outer pins. If you find continuity then something is wrong. Do the same for the pins of Q13, keeping in mind that pins 5 & 6 and pins 7 & 8 are intentionally connected together. If that still doesn’t help, check the board carefully for short circuits, especially between IC pins. Modifying the IMSC to run from 115V AC I purchased a couple of kits for your Induction Motor Speed Controller (April & May 2012; siliconchip.com. au/Series/25) way back when, and am now getting around to building them. With only a couple of minor mistakes, it has gone well. I was reading through your articles describing the design and function, and I have a couple of questions/requests. Most of this has to do with the fact that I live in the USA, and our mains power is 120V AC (240V AC for large appliances). I don’t want to have to use 240V AC for all my motor applications. What is the low-voltage cutoff? Is this a software feature that could be modified? I’d like to have the speed settings ratio be based on what we have for power here. That means a 60Hz default ramp up and a 90Hz top speed. I think that could also be changed in the software quite easily. Would you be willing to share the source code? I imagine I’m not the only one that would like to make a couple Australia’s electronics magazine of tweaks. GitHub and a public license to protect you and/or keep people from profiting from your work. (A. D., Columbia Heights, Minnesota USA) • You would need to change transformers T1 and T2 to run the IMSC from 110-120V AC as they would not produce a high enough output voltage otherwise. A possible alternative to changing these transformers would be to use the specified transformers, but change how they are wired to the diode bridges. T1’s configuration would need to change from a bridge rectifier to a full-wave voltage doubler, and T2 would need to be rewired to have its secondaries in series rather than in parallel. The under-voltage lock-out can’t be changed in software. If you modify T1/ T2 to produce the correct +15V HOT and 7V rail voltages with a ~115V AC input as described above, the circuit should operate normally. You couldn’t connect it to a 220-240V AC source after those modifications, though. The 0.5-75Hz range was chosen to include 60Hz as an option. The latest software gives a way to increase the maximum speed to 100Hz, which should be more than enough. We haven’t made the source code to this project freely available upon the designer’s request, because modifying it could be very dangerous. The IMSC is not something that inexperienced people should fiddle with, and we believe that giving away the source code would encourage that. Dual Power Supply wanted Has Silicon Chip ever designed a simple variable dual power supply? (R. M., Melville, WA) • Yes, we have published a few; the latest was the Dual Tracking Supply (June & July 2010; siliconchip.com.au/ Series/8). We have PCBs available for that project, and there is an Altronics kit (Cat K3218). If building that, please check our Notes & Errata page as there were some errata published for it. Other similar supplies published include: • Easy-to-Build Bench Power Supply (April 2002; siliconchip.com.au/ Article/4083) • Beginner’s Dual Rail Variable Power Supply (October 1994; siliconchip. com.au/Article/5220) continued on page 112 siliconchip.com.au • Dual Tracking ±50V Power Supply (April 1990; siliconchip.com.au/ Article/7258) • Dual Tracking ±18.5V Power Supply (January 1988; siliconchip.com.au/ Article/7828) the Jaycar Cat SY4080 (3A rated) and SY4084 (40A rated). These would need to be wired up and housed in an Earthed metal enclosure and wired according to the Australian wiring standards for mains equipment. Reducing switch wear from arcing Direct Injection Box query I have a computer (Apple Mac), a printer (Brother) and several other small items plugged into a powerboard fitted with a switch. After I have finished using the computer, I shut it down, wait until all the screen displays have switched off, then turn all the power off via the switch on the powerboard. Occasionally, there is a ‘blat’ sound that comes from the switch. I assume that this is a spark. I have had to replace the switch several times over the years, as the contact points in the switch have become stuck or welded together. Is there any way to reduce or eliminate this sparking? (G. H., via email) • One method to reduce switch contact wear due to arcing is to place an X2-rated 10nF 250V AC capacitor across the switch contacts (eg, Jaycar Cat RG5230). This will reduce the transient voltage across the switch contacts as they open. Adding the capacitor leaves a residual current flow that bypasses the open switch (around 8mA). Higher value capacitors can be used, and might suppress the sparking more effectively, but with a higher residual current. Another method is to switch the mains supply using an electronic switch such as a Triac. There are electronic relays that do this, such as Some years back, you presented an active direct injection box for guitars to plug into a PA system. The design included a low-cost transformer from Altronics or Jaycar and a JFET front end powered via the audio mixer phantom power supply. We built several of these for our local church and need to make more. While you can buy a commercial unit for around $100, I recall that these DI boxes were very cost-effective; certainly a lot less than $100. I can’t remember whether it was EA or Silicon Chip magazine. The DI boxes we constructed have proven to be very robust and deliver excellent sound quality. Can you advise when that project was published? (N. A., Canberra, ACT) • The DI Box design you are after is probably the one from Electronics Australia, February 1998 (97di12: “Direct Injection [active] Preamp using a JFET” ). You can order a scan of that article via www.siliconchip.com. au/Shop/15 Alternatively, Silicon Chip has published passive and active DI Boxes. Our passive version (October 2014; siliconchip.com.au/Article/8034) uses a high-quality transformer from Altronics, while the Active DI Box (August 2001; siliconchip.com.au/ Article/4158) does not use a transformer. SC Advertising Index Altronics...............................83-86 Ampec Technologies................. 49 Analog Devices........................... 7 Control Devices Australia............ 9 Dave Thompson...................... 111 Digi-Key Electronics.................... 3 Emona Instruments................. IBC Hare & Forbes............................. 5 Jaycar............................ IFC,53-60 Keith Rippon Kit Assembly...... 111 LD Electronics......................... 111 LEDsales................................. 111 Microchip Technology...... 13, OBC Ocean Controls........................... 6 SC Colour Maximite 2............... 75 Silicon Chip Binders............... 111 Silicon Chip Shop.............. 87, 98 Silicon Chip SiDRADIO............ 19 Switchmode Power Supplies..... 12 The Loudspeaker Kit.com......... 10 Tronixlabs................................ 111 Vintage Radio Repairs............ 111 Wagner Electronics................... 47 Weller Soldering Iron................. 11 Notes & Errata High-Current Battery Balancer, March 2021: in the parts list on p27, several Mosfets (Q11,Q12…) are listed as “S6M4” types. The correct type code is QS6M4. Arduino-based Adjustable Power Supply, February 2021: while the specified SY4030 relay from Jaycar is rated to carry 1A, it only has a 500mA switch rating. The similar S4100 relay from Altronics specifies a 1A switching current. Power supplies built using the Jaycar part should set the current limit no higher than 500mA to avoid damage to the relay. Other similar relays are available with a 1A contact rating; it appears that this refers to the carry current only, and not the switching current, so check the data sheet if substituting a different part. LED Party Strobe Mk2, August 2015: the link at lower-left should be positioned as shown in the photo on p87, not the overlay diagram (Fig.2) on p86, which incorrectly has it shown in the “MAX” position. The May 2021 issue is due on sale in newsagents by Thursday, April 29th. Expect postal delivery of subscription copies in Australia between April 27th and May 12th. 112 Silicon Chip Australia’s electronics magazine siliconchip.com.au