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 On Gerber files, FPGAs and PID coefficients I have just received my copy of the October issue in the mail to find that you plan a couple of things of particular interest to me. I like the idea of using a laser engraver to make a PCB, so I am looking forward to that article. Currently, I am looking at the Creality CP-01 from Altronics. But so far, I have not found a unit that accepts Gerber files, and thus the Gerber file has to be converted to JPEG, and then using a photo application to invert positive to negative. You have noted that you have not published many articles on FPGAs, and I think that is likely due to the difficulty in soldering these devices onto PCBs. Also, programming these devices requires some expensive tools. I would like to mention a possible improvement to the thermocouple wiring on DIY Solder Reflow Oven Controller (April-May 2020; siliconchip. com.au/Series/343). As you say, thermocouples are not that accurate, but it depends on the class. For class 1 it is ±0.5°C, and for classes 2 and 3, it is ±1°C. My suggestion is to obtain a type-K plug and panel-mount socket and use a short length of T/C cable cut off the thermocouple to go between the socket and T/C module. This eliminates the multiple cold junction points, which will not be at the same temperature due to the layout of the components. It also makes it just a bit more professional. Be aware that there may be a lacquer coating on the wire which needs to be cleaned off, and that the fibreglass insulation may make your hands itch. One thing on this unit is that the PID coefficients don’t seem to match what I know of standard PID settings, as the D value is so high. Is there a units change? Lastly, I’d like to comment on the Serviceman’s Log by Dave Thompson. I am glad he managed to fix his analog meter. It is lucky because the jewels in meters are usually spring-loaded and siliconchip.com.au so all that happened was one jewel was pushed down, enabling the other pivot to pop out. Simple to fix. What he could have had was one of the hairsprings having been bent so the coils touch each other, causing calibration problems or worse, a bent pointer. Both are fixable with great patience and care, but I won’t go into details on that. Please ask him if he has experience with taut band suspension meters such as the spot galvo. I’m not sure if the AVO was that or had standard pivot and jewels. (W. D. K., Bayswater, Vic) • You are right that you probably have to convert Gerber files to an image format for use with the laser engraver software. We cover all the steps in that article, which is planned for the December issue. Many FPGAs are available on development boards these days, removing soldering from the equation (many also come in quad flat packs which are not all that hard to solder with some practice). Software is becoming less of a problem too; many types now have free software available. We have covered these aspects in a couple of articles to date, and no doubt will have more to say in future. As for the thermocouple interface for the Solder Reflow Oven, we agree that adding proper thermocouple adapters is definitely nicer. But the cheaper and easier method described in those articles is adequate for the soldering task. There are substantial temperature differences around the oven, and a degree or two here and there in measurement is less than this variation. As for the PID parameters, the units used are not conventional; they are references to counts of the interrupt service routine (ISR). The PID loop parameters were tuned using an empirical approach. They are a compromise that gives good-but-not-ideal performance in the reflow oven application. Phil initially identified an appropriate value for P, which in a steady state gave reasonable behaviour. With I and Australia’s electronics magazine D at zero, there is a latency in settling that runs to many minutes. So he increased the I value to reduce the settling time. While this did not result in temperature oscillations, overshoot was observable, so the I value has been chosen to achieve a reasonable settling time with manageable overshoot. The D value was finally tweaked to reduce the overshoot. On considering your question, it might be that a reduction in I and a reduction in D would be better. Some lights do not meet electrical standards I have an oyster light fitting that can be clipped to a ceiling fan or be mounted as a ceiling light. There is no Earth connection at all to the metal base, and I am wondering whether this legal. I have been a subscriber of you magazine for years and are still enjoying it. (M. W., Murray Bridge, SA) • Such luminaires are not legal as the metal enclosure is not Earthed and there is no Earthing of the fitting if it has a metal connection for the lamp, such as a bayonet or Edison screw. According to the AS/NZ3000 wiring rules, “A protective earthing conductor, connected to a terminal or suitably insulated and enclosed, shall be provided at every lighting point. The exposed conductive parts of luminaires shall be earthed.” (section 5.4.3 Lighting points). Capacitors – more than meets the eye Have you published an article on capacitors? It seems there is more to these devices than meets the eye, and I am curious about why certain types are chosen for a specific application. This was initially triggered by the observation of the use of ceramics and green caps, then the rise and fall of tantalum caps. However, a couple of recent incidents have rekindled and broadened my curiosity. Recently, the circulation pump on November 2020  107 our solar hot water service became intermittent. The motor run capacitor was only a fraction of its rated capacitance when measured with a multimeter. Being out of town, I looked for a temporary replacement from my parts at hand. An old power supply had a couple of X2 capacitor rated at 250V and when connected in series, gave a little more than the required 0.8µF with the benefit of a theoretically increased voltage rating. This combination was bigger than the original but just fitted in the motor, and brought back reliable operation for three weeks until a proper motor run capacitor could be obtained. So what is the difference between the X2 caps and the motor run cap? The web page at siliconchip.com.au/ link/ab5k gave me confidence that X2s are suitable for mains use, while siliconchip.com.au/link/ab5l has a broader background which raises more questions. For example, how does one tell a polyester film and a polypropylene capacitor apart, given that they look similar? Finally, my son and I are trying to build an indicator light for the electric fence. We had the idea that a neon lamp in parallel with a capacitor that was charged via an 8kV, 500mA diode (UX-C2B) salvaged from a microwave power supply and a 51kW resistor might work as a “relaxation oscillator”. Testing it, it flashes every 3-4 electric fence pulses, and it does not load the fence energiser. However, I am now curious about the best capacitor to use. The trial used a 1µF polyester (or is it polypropylene?) rated at 250V from the same microwave switchmode power supply that donated the diode. (D. G., Koyuga, Vic) • Capacitors are indeed a topic with more depth than most people realise. Unfortunately, we haven’t published an in-depth article on capacitors, with the possible exception of our August 2002 article on tantalum capacitors (siliconchip.com.au/Article/6744). Your problem is prevalent – many motor failures are actually motor capacitor failures. These capacitors must have quite high capacitance and voltage ratings, especially motor start capacitors. Hence, start capacitors are usually some type of electrolytic, and they don’t tolerate long-term hightemperature operation well. According to Wikipedia (https://w. wiki/fJj), motor run capacitors are gen108 Silicon Chip erally polypropylene types as they must handle current continuously. Many X2 capacitors are also polypropylene, so provided they have a sufficient ripple current rating, they should be suitable. Some X2 capacitors are polyester, and those should be OK too, again as long as they can handle the current. There is no apparent difference in the appearance; you have to look up the part code to see if it is polyester or polypropylene. X2 capacitors are definitely suitable for mains use. The X part of the designation indicates that they are suitable for being connected between mains phases (eg, Active and Neutral). Y-class capacitors are suitable for connection between Active and Earth (they are required to fail opencircuit rather than short-circuit for safety) but can also safely be used between phases. X/Y-class capacitors can lose capacitance if abused (eg, exposed to high voltage spikes or passing more current than they are designed for). X/Y-class capacitors which are designed to handle significant currents, at least in the short-term, are sometimes referred to as “pulse” capacitors. We pulled up a data sheet for a randomly selected 2.2µF 275VAC X2 capacitor (Kemet R46KN422000P0M) to check its ratings. It is rated to handle just 250mA continuously at 50Hz, so it would only really be useful for a motor of about 60W. They make a 10µF version which is rated for more than 1A at 50Hz. So it appears that the main difference between a motor run capacitor and an X2 capacitor is that X2 capacitors are not designed to handle significant currents at mains frequencies, while motor run capacitors are. X2-class capacitors probably also have slightly different construction to meet their safety requirements. For your electric fence indicator, you want a low-leakage capacitor so ceramic, polyester or polypropylene should all be fine as long as they have a sufficiently high voltage rating. Presumably, that is limited by the neon as it will conduct at around 80V. For a proper discussion on capacitors, we would have to explain the many different ceramic dielectrics (NP0/C0G, X5R, X7R, Y5V etc), along with the many different plastic films used (polyester, polystyrene, PET, Australia’s electronics magazine polypropylene), varying plastic film construction methods, mica capacitors, electrolytic capacitors (aluminium, organic, solid, tantalum etc) and much more! Help fixing an Iamm Multimedia Player Years ago, I purchased an “Iamm HD Multimedia Player Cinema & Opera Juke Box” (model NTD36HD). This unit has never operated correctly. I tried to get it running after purchase, then put it aside and forgot about it. I found it again recently and thought what a waste it could not be used. I wondered if any of your staff or readers could be of any assistance, as I have had no success finding anything useful on the web. The hard drive is accessible via the USB socket and a computer. But when the unit is hooked up to a TV and audio system and powered up, it shows its start-up screen and plays its startup ‘music’, then very briefly goes to the screen displaying choices (movies, photos and music). The screen quickly goes blank, and the unit is effectively dead, save for the whirr of the harddrive still spinning. I have tried without success to find where I can get the ‘firmware’ to reload it. Does anyone out there have any experience with these multimedia players? (D. R., Goughs Bay, Vic) • We do not have any experience with that brand. Perhaps a reader can help. Linear Bench Supply voltage variations I am building the 45V 8A Linear Bench Supply from the October & November 2019 issues (siliconchip.com. au/Series/339). I have gotten to the point of the initial tests and calibrations before installing the main heatsink components, but some readings seem a bit off. The supply rails seem OK. The A5 pin of CON6 reads around 2.9V, which is under the 3-4V range suggested in the magazine, but the temperature reading when I plugged in the display matched close enough to a nearby thermometer. I did the initial calibrations so that TP5 measured exactly 15.6V and TP6 measured exactly 6V. TP1 and TP3 were both very close to 0V each. TP2 measures -115.7mV which is siliconchip.com.au below 0V as the article suggested, but TP4 is 12.1mV which is close to zero. I did have a little trouble fitting IC4 as I haven’t done any SMD soldering before, but after blasting it with hot air and clearing the bridges, I tested it in-situ and it seemed to be working. Is the 12.1mV reading anything to worry about? I have a little pocket oscilloscope which I used to test the oscillators. Pin 3 of IC3 was close to the 60kHz but about 51% duty cycle. The -5V rail was correct though. The Fan PWMs have me a bit worried. Pin 1 of IC2 measures 260Hz instead of the mentioned 280Hz, although it has an exact 50% duty cycle. However, pin 7 was showing a voltage. The displayed temperature was 30°C, so I carefully used an ice cube to drop the temperature back down to 25°C, and I still saw around 3V on pin 7. If it’s already on, does it need to drop lower than 25°C to switch off? (S. B., Banyo, Qld) • None of these readings concern us. The 12.1mV at TP4 corresponds to 16mA at the output, which is not precisely zero, but doesn’t sound excessive. It’s below the threshold of the meter readings. The PWM frequency isn’t critical; 260Hz is fine. We suspect that variation in the thermistor resistances and zener voltages mean your supply has a different temperature response. Compare the waveform on pin 5 of IC5 to the voltage on pin 6 (see the scope grabs on p28-29 in the October issue). If you see the duty cycle on pin 7 increase as the temperature increases, it is working correctly. You could try replacing the thermistor with a potentiometer (say 20kW, or at least above 10kW) and try sweeping it up and down to check the response. Using a Micromite as an audio scope Do you have software or do you intend to write a program for the Micromite Backpack to be an X-Y vector scope? I need to display the Lissajous pattern of a stereo audio signal without tying up my CRO. (P. S., Mount Pleasant, SA) • Peter Mather has posted a twochannel timebase scope CFUNCTION on the Back Shed Forum at www. thebackshed.com/forum/ViewTopic. php?TID=8077 siliconchip.com.au It has a ~1MHz sample rate. He has also posted the source code. A quick glance through it suggests that the code which draws the pixels as X/T and Y/T could be combined to draw X/Y. How RMS power is determined Thank you for your wonderful magazine, which I have been purchasing and reading since 1987. Concerning the Ultra-LD Mk4 amplifier project (August-October 2015; siliconchip.com.au/Series/289), the rated power is listed at 135W RMS into 8W with ±57V DC supply rails. The term RMS is generally used to refer to voltage or current and not power. Power (in the audio industry) is simply the product of RMS volts and RMS amps where the signal is a sinusoidal wave. With supply voltages of ±57V DC, my calculations show that the maximum RMS voltage is 40.3V RMS (57 ÷ √2). Hence the power would be 203W RMS (V2 ÷ R) into a resistive load of 8W. Even with a Vce(sat) max of 3V for the output transistors used, it is difficult for me to see how the 135V RMS is derived. Are you able to shed alight as to why the power rating is 135V RMS for this superb amplifier? (J. D. S., Endeavour Hills, Vic) • You are right that the term “RMS power” is confusing, but it is common. As described in Wikipedia at the following link, “RMS power” is the power measured or calculated with a continuous sinusoidal signal (ie, it’s calculated based on the RMS sinewave voltage): siliconchip.com. au/link/ab4o We measure it by increasing the signal level until the point where distortion starts to rise, then measuring the continuous power delivered at that setting. Your calculation ignores several important factors such as the fact that the output voltage cannot swing railto-rail (due to several factors, including the driver and output transistor base-emitter voltages). Plus the supply voltage will not remain at ±57V DC at full load, and there will be significant ripple on the supply rails, which will lead to earlier clipping. There are also losses in the output transistors (as you point out), losses Australia’s electronics magazine in the output filter, losses in the wiring and tracks etc. In short, you have to measure the real-world power delivery (or a very accurate simulation). The music power is stated as being somewhat higher than 135W as this is a short-term measurement and so the supply voltage will not sag as badly. You could probably get 150W RMS from this amplifier module, or perhaps a little bit more, with a larger transformer and larger supply filter capacitors which would both help to reduce supply voltage sag and ripple under load. How much do precision voltage references drift? On many occasions, I have appreciated the value of the Simplified 10V Precision Voltage Reference by Jim Rowe (August 2014; siliconchip.com. au/Article/7976). The IC is now six years old, and my version still runs happily on its original 9V batteries. My question is: how significant is the age-related degradation that has taken place? Also, what would be the best and/ or most economical method to recalibrate it if necessary? Is there a better or more accurate standard easily achievable? Thanks for the great magazine, keep up the good work. (C. O. D., Adelaide, SA) • Jim Rowe responds: It’s good to hear that you have found the Precision Voltage Reference of use. Analog Devices quote the ageing rate of the AD587 device as ±15ppm per 1000 hours, but this figure of 1000 hours probably refers to hours of operation rather than merely the passage of time. So unless you have been using your Voltage Reference continuously over the last six years, I would expect that it would still be very close to its original calibration. As my original prototype has only been used about two or three times a year in the last six years, I thought I would test it this morning with three different reference instruments. The readings I obtained were 10.001V, 9.999V and 9.9976V – with the last figure from a Yokogawa 7562 bench DMM which has itself not been recalibrated since 2010. The testing was done at 16.3°C, about 9°C cooler than the original testing temperature in 2014. This suggests that your Reference is probably still November 2020  109 quite accurate too, which is good news, since it isn’t all that easy to recalibrate. Adjusting Mosfet dead time with a scope Is there any way to set the dead time on the Class-D amplifier module (November & December 2012; siliconchip. com.au/Series/17) using an oscilloscope? (B. C., Albion, Vic) • It would be possible to use a scope to observe the Mosfets switching on and off to help guide you in setting the dead time to the optimal value. But doing so is quite tricky as the upper Mosfet in each pair is ‘floating’, so measuring their gate-source voltages would require an isolated probe, or a scope with individually isolated channels. An easier approach would be to insert a shunt in the ground connection of each pair of Mosfets and monitor the voltage across it. The dead time setting is optimal when it is set as short as possible without a large spike in current draw during the transition period, when one Mosfet switches off and the other switches on. You would need to break the track and solder in a shunt, and given that its value would need to be low, you’d need a pretty sensitive scope or amplifier. But it could be done. And this would have the distinct advantage that it would take into account the switch-on and switch-off delays of each Mosfet, which cannot be determined by merely observing their gate drive waveforms. Output from photoelectric smoke alarms Around 20 years ago, I built your Smoke Alarm Control Panel project (January & February 1997; siliconchip. com.au/Series/149). The installation has been running since then without any hiccups, bar the replacement of a couple of ICs and the power supply. I check it annually. A while back, I decided to replace the aging Kambrook smoke detectors with newer Quell detectors, namely the Q946 ionisation-type detectors. These appear to use an A5364CA CMOS IC. The replacement was a fairly straightforward exercise, and the new detectors work as intended. The problem I now have is that Quell made available a different smoke 110 Silicon Chip detector for the kitchen location, which is a photoelectric type (Q301H). I cannot figure out how to interconnect the alarm output from this device to the Control Panel. I would be grateful if you can provide me with advice and help with this issue, as I do not want to have the kitchen area unprotected. (H. B., Mt Kuring-Gai, NSW) • The alarm output from the photoelectric smoke detector (Q301H) should be available at pin 10 of the A5364CA IC. For the alarm test input, use the additional circuit of the Control Panel for Smoke Alarms comprising Q4, except using a 200kW resistor (instead of the 1MW resistor) at Q4’s collector. Connect the opposite end of the 200kW resistor directly to the “push to test” button. Stopping nuisance smoke alarms Do you know of a clever way I can turn off the smoke detector while cooking? Also, have you designed an aspect ratio converter? I need to convert VHS footage from 4x3 to 16x9 without using a computer. (J. H., via email) • We published a smoke detector kill switch in February 1996 to prevent an alarm when cooking: siliconchip.com. au/Article/5038 We have not published an aspect ratio converter. There are commercially available units such as the Miranda ARC371P; we suggest you try one of those: siliconchip.com.au/link/ab5j Building a sinewave inverter I am wondering if you have a design for a pure sinewave inverter (230V AC). (A. R., Eltham, Vic) • We published a 2kW 24V DC to 230V AC pure sinewave inverter in the October 1992 to February 1993 issues (siliconchip.com.au/Series/173). That design is outdated, but we have not updated it, since commercial versions are far cheaper now. There is no way we could design an inverter for the cost that you could buy one these days. Combining AND gates for a clock I want to build a clock with local time, UTC and sidereal time on six 7-segment displays. I would like the Australia’s electronics magazine same crystal to run all the clocks. I have found a circuit to generate the 1.002738Hz for the sidereal clock, and it also generates a 50Hz for the local time. But it requires a 1MHz crystal and a 4068B IC (8-input AND gate). I found out that the 4060B IC can do frequency division and could be capable of dividing 2MHz to 1MHz. The 4068B IC is now hard to find (Mouser has it, but the delivery cost is prohibitive). Can I use seven 2-input AND gates instead? And how can I divide the 50Hz signal down to 1Hz for the local clock? (R. M., Melville, WA) • Yes, you can make up an 8-input AND gate from seven cascaded 2-input AND gates. A 50Hz to 1Hz divider circuit is shown at siliconchip.com. au/link/ab5i It uses two 4017B ICs fed with the 50Hz signal from the secondary of a mains transformer, but you could feed in the 50Hz output from your digital divider instead. Looking for historical documents I worked at Fairchild Australia in Melbourne from 1965 to 1974 as an Applications Engineer and Manufacturing Manager. Recently, I was asked to contribute to a history of the manufacture of semiconductors in Australia. I remember Jamieson Rowe visited the Fairchild factory at Kilsyth and then wrote an article about our factory. I think Electronics Australia also published articles on other manufacturing facilities over the years. I would be very grateful if someone could point me in the direction of any such articles that I could use. Do you have a listing of all articles that I could scan? I am interested in the period from the 1950s to the 1980s. (B. O. S., Blackburn, Vic) • Jim Rowe responds: after a bit of searching back through old EA indices, I believe I have found that article to which you are referring. It was in the February 1973 issue, and titled “Fairchild now making TO-92 transistors here”. The only other articles on Australia’s short-lived semiconductor industry I came across were these: May 1972: “Local Semiconductor Breakthrough” June 1972: “Centre Industries Making GE Diodes” March 1973: “Philips’ Hendon facility in SA” SC siliconchip.com.au