Hi all,
While repairing equipment, I sometimes find design errors in commercial equipment. (That is, a cynic might say the circuits are designed to fail after the warranty expires, but I take the optimistic view that they are errors.) Please add your own known design errors to the thread.
Philips FR675:
Each channel has a 2SC2240 transistor that is used as a current source of about 9 mA to 10 mA having almost 50 V across it. That gives a power dissipation of some 400 mW to 500 mW, while the 2SC2240 datasheet states an absolute maximum power of 300 mW at 25 degrees C. The 2SA970 that is used as a degenerated common-emitter stage has the exact same problem: 400 mW to 500 mW in a 300 mW transistor.
Technics SU-VZ320:
The freewheeling diode across the 24 V relay coil is a Panasonic MA165, which is rated for 35 V. The relay coil is supplied from an unregulated -42 V supply via resistors that form a voltage divider with the relay coil's resistance. Due to the relay coil's inductance, the poor old 35 V diode is subjected to the full 42 V for a few milliseconds during turn on.
Early versions of the DATEQ BCS200 mixing console:
The gates of the P-channel JFETs that select the audio for the monitors are pulled up via indicator lamps. When a lamp burns out or loses contact, the JFETs can turn on and off randomly depending on PCB leakage current. It has been fixed in later models by connecting resistors in parallel with the lamps.
The resistors in the second-order 20 Hz/80 Hz Sallen and Key high-pass filters in early versions of the DATEQ BCS200 mixing console are swapped, resulting in a very low Q and very gradual roll-off. It is corrected in later models.
Sony TA1150:
As shown in this thread: https://www.diyaudio.com/community/...am-unsure-how-to-proceed.412059/#post-7667649 , the bias trimming potmeters are connected such that the quiescent current goes through the roof instead of to zero when the wiper loses contact.
Regards,
Marcel
While repairing equipment, I sometimes find design errors in commercial equipment. (That is, a cynic might say the circuits are designed to fail after the warranty expires, but I take the optimistic view that they are errors.) Please add your own known design errors to the thread.
Philips FR675:
Each channel has a 2SC2240 transistor that is used as a current source of about 9 mA to 10 mA having almost 50 V across it. That gives a power dissipation of some 400 mW to 500 mW, while the 2SC2240 datasheet states an absolute maximum power of 300 mW at 25 degrees C. The 2SA970 that is used as a degenerated common-emitter stage has the exact same problem: 400 mW to 500 mW in a 300 mW transistor.
Technics SU-VZ320:
The freewheeling diode across the 24 V relay coil is a Panasonic MA165, which is rated for 35 V. The relay coil is supplied from an unregulated -42 V supply via resistors that form a voltage divider with the relay coil's resistance. Due to the relay coil's inductance, the poor old 35 V diode is subjected to the full 42 V for a few milliseconds during turn on.
Early versions of the DATEQ BCS200 mixing console:
The gates of the P-channel JFETs that select the audio for the monitors are pulled up via indicator lamps. When a lamp burns out or loses contact, the JFETs can turn on and off randomly depending on PCB leakage current. It has been fixed in later models by connecting resistors in parallel with the lamps.
The resistors in the second-order 20 Hz/80 Hz Sallen and Key high-pass filters in early versions of the DATEQ BCS200 mixing console are swapped, resulting in a very low Q and very gradual roll-off. It is corrected in later models.
Sony TA1150:
As shown in this thread: https://www.diyaudio.com/community/...am-unsure-how-to-proceed.412059/#post-7667649 , the bias trimming potmeters are connected such that the quiescent current goes through the roof instead of to zero when the wiper loses contact.
Regards,
Marcel
Last edited:
Many electrolytes placed wrong (polarity reversed) on low voltage applications ("They will survive.").
Simplified grounding by the marketing department for cosmetic reasons only (countless failures).
Applying a 2SK42 jfet rated for 15V in a near 30V amplifier section running at 60µA. It does survive!
A N-ch enhencement mosfet, switching on and off batteries (3xAAA), current running from source to drain, driven with a normal single pole switch operated by the user, shorting gate to source ('off') or leaving the gate floating unconnected ('on').
Keyboard matrix wiring extended up to over 10m on wireless transmitters.
High capacity supply rail electrolytes fixed to hot heat sinks.
Simplified grounding by the marketing department for cosmetic reasons only (countless failures).
Applying a 2SK42 jfet rated for 15V in a near 30V amplifier section running at 60µA. It does survive!
A N-ch enhencement mosfet, switching on and off batteries (3xAAA), current running from source to drain, driven with a normal single pole switch operated by the user, shorting gate to source ('off') or leaving the gate floating unconnected ('on').
Keyboard matrix wiring extended up to over 10m on wireless transmitters.
High capacity supply rail electrolytes fixed to hot heat sinks.
1. A very long time ago, Yamaha P2200 amplifiers had a bad rail sticking and recovery problem on the positive side that sound men blamed for failing horns/ high drivers. The best I found at the time was clamping diodes on the LTP collectors, but I think I could do better today. That amp also had VI protection that reacted very badly with a (~70V) speaker transformer. I think a large cap fixed the problem.
2. I fixed an airplane icing detector that had a relay with no coil damping, no diode, and the relay driver transistor had been blown. And it had other problems too.
3. A roller rink had maybe a speaker wiring short to ground that caused ground loop problems, oscillations when the system was turned up. Not sure. But I added small caps to the output relay circuit so that 10KHz+ at power disconnected the speakers and it never failed again.
4. A tape player would flutter. It had a negative resistance speed regulator where the current sense resistor was the wrong value.
5. A ~"Soundcraft" mixer hummed, because the ground bus was physically separated from the mixing bus wire(s). Adding/moving a ground wire across the cards next to the mixing bus wire fixed it, reduced the loop area.
6. We sold a TRIAC control where one channel was used for a confetti cannon. The solenoid would inductive kick-back and keep the triac on. An MOV fixed the problem.
7. Electricians wired a duplex wall socket at the back of a ~church using the neutral on a GFI breaker for the sound system. So when someone plugged something into that power outlet, down went the sound system.
8. A hotel had a huge power conduit in the parking lot, under the events rooms. The conduit was aluminum and the sound system hummed. They added a steel cover on the conduit.
9. A Bogen tube PA amplifier had asymmetric clipping. I changed a resistor for the ~12Ax7 driver circuit so it didn't clip prematurely, and it went from 60W to 100W output.
10. We had some (quasi) amplifier modules that hummed when you pushed the speaker cone in one direction because the Sziklai side was unstable. It needed a Zobel and build-out inductor + resistor.
11. There is/was a home intercom company that sold "magic" expensive 3-wire lamp cord for their product. The magic is/was that the outside conductors are the input and output of the amplifier and need to be shielded from each other. Center conductor being ground.
nuff 4 now
2. I fixed an airplane icing detector that had a relay with no coil damping, no diode, and the relay driver transistor had been blown. And it had other problems too.
3. A roller rink had maybe a speaker wiring short to ground that caused ground loop problems, oscillations when the system was turned up. Not sure. But I added small caps to the output relay circuit so that 10KHz+ at power disconnected the speakers and it never failed again.
4. A tape player would flutter. It had a negative resistance speed regulator where the current sense resistor was the wrong value.
5. A ~"Soundcraft" mixer hummed, because the ground bus was physically separated from the mixing bus wire(s). Adding/moving a ground wire across the cards next to the mixing bus wire fixed it, reduced the loop area.
6. We sold a TRIAC control where one channel was used for a confetti cannon. The solenoid would inductive kick-back and keep the triac on. An MOV fixed the problem.
7. Electricians wired a duplex wall socket at the back of a ~church using the neutral on a GFI breaker for the sound system. So when someone plugged something into that power outlet, down went the sound system.
8. A hotel had a huge power conduit in the parking lot, under the events rooms. The conduit was aluminum and the sound system hummed. They added a steel cover on the conduit.
9. A Bogen tube PA amplifier had asymmetric clipping. I changed a resistor for the ~12Ax7 driver circuit so it didn't clip prematurely, and it went from 60W to 100W output.
10. We had some (quasi) amplifier modules that hummed when you pushed the speaker cone in one direction because the Sziklai side was unstable. It needed a Zobel and build-out inductor + resistor.
11. There is/was a home intercom company that sold "magic" expensive 3-wire lamp cord for their product. The magic is/was that the outside conductors are the input and output of the amplifier and need to be shielded from each other. Center conductor being ground.
nuff 4 now
As I was repairman and tech of various kinds and also am of the cynic kind I have seen many of these and it was kind of sport to solve them. Later on one becomes experienced in solving errors before the failure occurs which is most optimal IMHO. Today we are seeing planned obsolescence and one does not need to be a cynic to detect them. For instance Braun and Dyson products have these in abundance. Dyson with 2 kW motors switched on with too light switches and no inrush protection thereby burning the carbon brushes and switches in no time, Braun for making plastic extra thin like the cover on KF47 or for instance the electric toothbrushes that could easily be repaired but are made to be opened only once after production just to throw them away. Lately I saw a washing dryer that had no bearing but a plastic ball turning in metal taking all the forces. Bosch/Siemens dish washers that have a flexiprint of the heatpump that becomes open circuit as calcium will stick to the flexiprint making it a hot spot causing the flexipint track to melt. Akai devices that had no decoupling caps at all and this was sold as a feature. Akai tuners full class A that needed a modification by means of a heatsink for the transformer. In several brands regulators regulating let's say +/- 50V to +/- 12V for opamps burning several Watts and no heatsink around of course. Once one maneuvers in industry of consumer goods one gets an idea here and there when talking to designers. Many if not most products really are designed to last 2 years, plain and simple. Modern audio is full of design imperfections too many to mention really, even white goods have less of those. So far for the green dream 
In industry I have seen absolutely unbelievable errors made by "true professionals" either by ignorance, lack of time, unrealistic deadlines or sheer Dunning Kruger. What about "DC transformers" for photovoltaic installations? 5 kV DC square waves on a 3 phase transformer and MW power. It worked ... for a while. One of the better ones was a UPS system for a country far away with galvanic isolation so transformers of very high power on input, bypass and output... wound for the wrong mains voltage. To make it even better this was done in 2 identical systems. Coincidentally I was the guy that mentioned the mains voltage of the country they were meant to be shipped to. Yeah that made me a popular guy
In industry I have seen absolutely unbelievable errors made by "true professionals" either by ignorance, lack of time, unrealistic deadlines or sheer Dunning Kruger. What about "DC transformers" for photovoltaic installations? 5 kV DC square waves on a 3 phase transformer and MW power. It worked ... for a while. One of the better ones was a UPS system for a country far away with galvanic isolation so transformers of very high power on input, bypass and output... wound for the wrong mains voltage. To make it even better this was done in 2 identical systems. Coincidentally I was the guy that mentioned the mains voltage of the country they were meant to be shipped to. Yeah that made me a popular guy
Last edited:
A contribute to low reliability and poor reparability of some of the latest electronic devices is the over reliance on software-based functions. There is the idea that the only way to get to the result is to use a microcontroller, even for trivial fixed-function devices that are best built with hardwired logic. A cheap microcontroller may cost the same, but is introducing several failure modes such as increased susceptibility to EMI, software errors, and at least one extra manufacturing step (loading the software on the microcontroller). The mayor advantages of the microcontroller-based design are BOM simplification (same parts are used over a large number of products) and reduced testing time (no need to pour time and effort on analog design or custom logic). But the end result is the "let's try to unplug and turn on again to fix it" syndrome that is sometimes plain silly.
Yes although of the electric/electronic persuasion one eventually learns the merits of simplicity and “one function only” stuff like real mains switches, real 50/60 Hz transformers, mechanical sturdyness etc.
Stuff that only needs to do 1 thing tends to do that rather well. Just read a technical review of a streamer that can do many things via all possible interfaces and can be controlled by app and beautiful touch buttons except that it can not play back music a good way. It listens to Alexa though.
Form follows function or is it the other way around?!
Stuff that only needs to do 1 thing tends to do that rather well. Just read a technical review of a streamer that can do many things via all possible interfaces and can be controlled by app and beautiful touch buttons except that it can not play back music a good way. It listens to Alexa though.
Form follows function or is it the other way around?!
Last edited:
Krell KAV300i has its low voltage rails made with a simple linear reg with zeners for reference and those things run so hot that not only the resistors burn, but all the capacitors nearby too.
The brown ones used to be nice and orange in distant past lol.
I still have to figure out why the amp is misbehaving, its final output stage is outside the feedback loop and there's rising freq response with volume level, and both channels differ. Unfortunately the layout inside makes detective work a bit difficult. One day I'll have enough desk space to have it open and under close scrutiny...
The brown ones used to be nice and orange in distant past lol.
I still have to figure out why the amp is misbehaving, its final output stage is outside the feedback loop and there's rising freq response with volume level, and both channels differ. Unfortunately the layout inside makes detective work a bit difficult. One day I'll have enough desk space to have it open and under close scrutiny...
Attachments
I have seen innumerable design flaws, and I don't think any of them was deliberate: it was just plain, simple incompetence.
One or two examples: in the display modules of petrol pumps that had been upgraded from incandescent Numitron types to fluorescent displays, a higher voltage than the available 5 & 12V supplies was required.
To create the voltage, a voltage multiplier driven by power darlingtons was used, and to regulate the voltage, the µC width-modulated the driving squarewaves. Problem, there was no inductor anywhere, and the theoretical output voltage of a multiplier only depends on the peak voltage, not the conduction time.
It sort of worked though, because smaller duty cycles meant increased losses, which lowered the voltage, but it put an immense stress on the capacitors, diodes and transistors, and the scheme was in fact an informal linear regulator, with the esr of the caps, the tracks, the active components playing the role of a ballast.
Needless to say that these modules had severe reliability issues....
An Israeli telecom manufacturer had an unbeatable CWDM system on offer: it was much cheaper than the competitors. Of course, we wanted to evaluate it, but when an OSA was connected to the output, nothing was visible, yet it seemed to work. Quite puzzling, but digging deeper we found that the temperature of the lasers (critical for the wavelength accuracy) was handled by a bang-bang controller: the wavelength was all over the place, but the average was correct...
Another well-known telecom manufacturer had a new line of PDH products. The stability, BER, etc were atrocious. I found that they had used CMOS gates in linear mode; why not after all. Problem, the system operated at 34MHz and they used AC types for their speed. AC gates are always buffered and have intrinsic hysteresis on all inputs: a recipe for instability and disaster..
One or two examples: in the display modules of petrol pumps that had been upgraded from incandescent Numitron types to fluorescent displays, a higher voltage than the available 5 & 12V supplies was required.
To create the voltage, a voltage multiplier driven by power darlingtons was used, and to regulate the voltage, the µC width-modulated the driving squarewaves. Problem, there was no inductor anywhere, and the theoretical output voltage of a multiplier only depends on the peak voltage, not the conduction time.
It sort of worked though, because smaller duty cycles meant increased losses, which lowered the voltage, but it put an immense stress on the capacitors, diodes and transistors, and the scheme was in fact an informal linear regulator, with the esr of the caps, the tracks, the active components playing the role of a ballast.
Needless to say that these modules had severe reliability issues....
An Israeli telecom manufacturer had an unbeatable CWDM system on offer: it was much cheaper than the competitors. Of course, we wanted to evaluate it, but when an OSA was connected to the output, nothing was visible, yet it seemed to work. Quite puzzling, but digging deeper we found that the temperature of the lasers (critical for the wavelength accuracy) was handled by a bang-bang controller: the wavelength was all over the place, but the average was correct...
Another well-known telecom manufacturer had a new line of PDH products. The stability, BER, etc were atrocious. I found that they had used CMOS gates in linear mode; why not after all. Problem, the system operated at 34MHz and they used AC types for their speed. AC gates are always buffered and have intrinsic hysteresis on all inputs: a recipe for instability and disaster..
Microcontrollers are indeed used all over the place, I've seen vacuum cleaner speed controls where the resistor, capacitor and DIAC were replaced by a microcontroller.
Over the decades in the repair shop, I've seen plenty of mistakes and/or stupid designs too.
Off the top of my head I can't remember anything specific, but I do remember shaking my head in disgust over some designs.
Some were due to cost-cutting, others were poor layout or design.
You have to remember, those things were initially designed by humans, however some of the newer crap is because of humans using 'apps' which are just as questionable as a human's brain.
After all, 'apps' are designed by imperfect humans, right?
Off the top of my head I can't remember anything specific, but I do remember shaking my head in disgust over some designs.
Some were due to cost-cutting, others were poor layout or design.
You have to remember, those things were initially designed by humans, however some of the newer crap is because of humans using 'apps' which are just as questionable as a human's brain.
After all, 'apps' are designed by imperfect humans, right?
Three generic things come to mind:
1) Big voltage dropping resistors directly soldered to filter cap lugs (typically tube equipment) so the heat gets conducted right into the capacitor, shortening the life. Pretty much everybody did this.
2) Filter caps with lugs supported or tensioned against the lugs, risking cracking the top of the (expensive) caps. Crown did this in a couple cases.
3) Lack of understanding of how to connect grounds, mixing AC charging currents into low level signals. Not sure everybody gets it even today.
1) Big voltage dropping resistors directly soldered to filter cap lugs (typically tube equipment) so the heat gets conducted right into the capacitor, shortening the life. Pretty much everybody did this.
2) Filter caps with lugs supported or tensioned against the lugs, risking cracking the top of the (expensive) caps. Crown did this in a couple cases.
3) Lack of understanding of how to connect grounds, mixing AC charging currents into low level signals. Not sure everybody gets it even today.
Well, I don't do a lot of SS repair, but 2 come to mind:
A battery-only stair climber control board that had a big current draw due to a TIPxx with a fairly large heatsink. As it would need a new set of batteries every workout AND the battery door could only be removed by unbolting the entire control unit, I got it free from a friend and put a proper external PS port on it.
The 1st gen eero router had a surface mount barrel jack connector held on only by the solder pads. Got one cheap and solder fixed it right up. 2nd gen onwards used USB-C connectors.
A battery-only stair climber control board that had a big current draw due to a TIPxx with a fairly large heatsink. As it would need a new set of batteries every workout AND the battery door could only be removed by unbolting the entire control unit, I got it free from a friend and put a proper external PS port on it.
The 1st gen eero router had a surface mount barrel jack connector held on only by the solder pads. Got one cheap and solder fixed it right up. 2nd gen onwards used USB-C connectors.
Last edited:
Years ago I did some electrical work an a friends 2010 Nissan Sentra.
The fuse box in the engine compartment was rediculous.
First remove the plastic cover - normal.
Then lift out and flip over a large assembly that held fuses and relays, upside-down - not normal.
Of course the attached wire bundle was a bit springy and always wanted to flip it back.
Now the molded black-on-black text for fuse identifcation was also upside-down making it even harder to read.
And finally, the fuse that needed replacement was in the "other" fuse box, hidden under the dash between the steering wheel and the outside wall.
Inaccessable to nearly everyone except perhaps small children and really short midgets.
I suppose "easy of service" was not a design consideration at the time.
The fuse box in the engine compartment was rediculous.
First remove the plastic cover - normal.
Then lift out and flip over a large assembly that held fuses and relays, upside-down - not normal.
Of course the attached wire bundle was a bit springy and always wanted to flip it back.
Now the molded black-on-black text for fuse identifcation was also upside-down making it even harder to read.
And finally, the fuse that needed replacement was in the "other" fuse box, hidden under the dash between the steering wheel and the outside wall.
Inaccessable to nearly everyone except perhaps small children and really short midgets.
I suppose "easy of service" was not a design consideration at the time.
Back to audio gear: using closed (nearly air tight) enclosures for amps, especially power amps, is a design error.
Not only is the circulation of air and associated cooling prevented, any air humidity will remain in the case and
cycle between vaporisation and condensation causing unnecessary corrosion. Transpiring gases from rubber and
other materials quickly increase this corrosion and decrease usability time by a large margin in some amp models.
All electronic gear encapsulation should have holes, vents, slot openings or whatever you want to use or call it.
It can also be the other way round: an 8-tube preamp of upper price class in all-closed cabinet was able to soak
all the smoke from cigarettes it could gather through some only hair wide openings between aluminium sheets
until the inside was brown/black. I can supply pictures of this ..
Not only is the circulation of air and associated cooling prevented, any air humidity will remain in the case and
cycle between vaporisation and condensation causing unnecessary corrosion. Transpiring gases from rubber and
other materials quickly increase this corrosion and decrease usability time by a large margin in some amp models.
All electronic gear encapsulation should have holes, vents, slot openings or whatever you want to use or call it.
It can also be the other way round: an 8-tube preamp of upper price class in all-closed cabinet was able to soak
all the smoke from cigarettes it could gather through some only hair wide openings between aluminium sheets
until the inside was brown/black. I can supply pictures of this ..