You are here

D.I.Y. amp

Page 6 of 7

Re: D.I.Y. amp

PostPosted: Tue Mar 12, 2019 5:10 pm
by blinddrew
Bummer :(

Re: D.I.Y. amp

PostPosted: Tue Mar 12, 2019 6:04 pm
by ef37a
Folderol wrote:Well today with the decks cleared, and after extensive testing of the new module on its pcb I carefully connected it up to the sub amp, checked all connections multiple times and switched on.

... I've just put an order in to RS for 28 quids worth of power transistors :madas:

Did you not manage to liberate a Variac when you retired Will? And do NOT let anyone buy you Duggie Self's amp book, it will just depress you further!


Re: D.I.Y. amp

PostPosted: Wed Mar 13, 2019 3:32 pm
by Folderol
The only variac they have is a massive lump that takes two people to lift - not exactly the sort of thing you can hide in your pocket. Besides, I wasn't expecting any kind of problem. The power stage has run for several years without issues and the new module tested out fine under quite extreme conditions. I still don't know what caused the problem, so am being mega cautious with further tests.

Anyway, wound-licking over and new transistors arrived, so ...

When I removed the remains of the two output transistors I discovered that the other three transistors in the power section escaped unscathed - so at least that part of the protection system I designed in had worked. Just to be sure I applied full voltage to them and controlled drive current {i.e. a resistor}. All three passed. The slight downside is that I spent more than I necessary.

Something that may be a problem is that although I bought extra output ones, I can't get anywhere near a matched pair. Frustratingly all the NPN ones have a gain close to 50, and the PNP ones around 110. That's most unusual. We'll see how that pans out on a test rig.

Re: D.I.Y. amp

PostPosted: Sat Mar 16, 2019 6:03 pm
by Folderol
Well it's back to the drawing board I'm afraid :cry:

The good news is that the amp is now back as it was, and seemingly none the worse for the 'experience' apart from some scorch marks. I gave it a good blast for about an hour this afternoon - actually playing some music for a change :tongue:

However, tacking a spare pair of those transistors onto the new module and running under very controlled conditions immediately revealed the cause of the failure. Massive parasitic oscillations at a frequency that's so high a 10MHz scope can barely resolve them :o Under these conditions both transistors stay almost permanently 'on', effectively a short circuit across the supply.

I remember an old college professor once saying that the correct way to deal with these is not to cause them in the first place! With such a bad case I didn't even bother to try the usual dodges. Instead, I went over the board and design with a fine tooth comb to see if I could find any stupid errors - couldn't see any. My best guess is that the extra stage pushed the open-loop gain up too high - combined with the fact that modern transistors have an incredibly wide bandwidth. Something I forgot was that coupling the module to the output stage also increases to module gain as it no longer directly takes the feedback resistors.

I'll give this a break for a while, and maybe get on with the front panel instead.

Re: D.I.Y. amp

PostPosted: Sun Aug 04, 2019 11:15 am
by Folderol
You were expecting me to report I had finally made the front panel... sorry :(

So no, not really any progress - be warned. Don't ever retire - you never seem to have time to do anything!

Anyway the other day (for some undefined value of 'other') there was an incident :o
I was happily playing some bit of improvisation when suddenly all sound cut out. The monitor was showing everything seemingly quite normal, but then I glanced at the amp display, to be greeted by both channels showing an offset of around -2V and wavering. The good news is that the monitoring system had done exactly what it was designed for and disconnected the speakers.

Puzzled, I switched on the Sub channel, which showed exactly the same behaviour, so this was clearly a common fault. Pulling out the input plugs proved that it was in no way an external issue, and the voltage started to quite rapidly go even more negative, and before i could reach the power switch suddenly slammed to full positive rail.

Today, withdrawal symptoms became so severe I dropped everything to investigate the amp. Cold checks revealed the internal fuse for the negative rail to the voltage regulators had blown. This is rated at 100mA, although only about 50mA should be drawn.

Here is the drawing for it. I was a bit relieved that this suggested it was unlikely to be anything to do with the (expensive) power stages, and initially thought it might be the -12V regulator or associated caps, but cold checks showed nothing suspicious.

My next move was to pull the other fuse, then power just this stage from my bench PSU, with current limiting set to 100mA. This immediately limited at 100mA for the negative supply and dropped the voltage to (curiously) around -12V. However, the output of that regulator was around - 9V, so it clearly wasn't shorting, and unlikely to be the problem.
These regulators feed the preamp stages, so I tried disconnecting them - no difference.

This suggested a problem with the -20V rail which is used by the drive stage of the amp itself, so I disconnected the feeds here and the other supplies all came up correctly, with both rails drawing about 53mA.

Cold checks seemed to show no problems with the PCB components, which includes everything up to the four power transistors for each output stage, and without the main supplies these just sit passively at 0V.

Finally I started to make live voltage checks. There really is only one common point, and that is the BC560 bias transistor (on the PSU drawing). This is mounted on the PCB in close proximity to the input transistors so that they are all at a similar temperature. I quickly discovered the collector was sitting at about -12V i.e. the reduced negative rail.

My first though was that the 100u cap might have gone S/C this cutting off the transistor, but that also had about 12V across it! Thinking that possibly the transistor had gone O/C (very unlikely) I switched off and did cold checks on it, getting perfectly correct readings.

Time for a cup of tea!

I must be getting {cough} old {cough} because it took me quite a while for the obvious to sink in. If the transistor was OK the voltage couldn't possibly be wrong... unless it was disconnected, and continuity testing showed that the base was, indeed, waving about in the breeze.

Confession time

All those components are physically on the same PCB track, so only a bad joint could be the cause. I shall now go and stand in the corner wearing a dunce's hat :oops:

Re: D.I.Y. amp

PostPosted: Sun Aug 04, 2019 12:27 pm
by blinddrew
Just put it down to insufficient cups of tea...

Re: D.I.Y. amp

PostPosted: Fri Jun 12, 2020 3:52 pm
by Folderol
Anyone remember my abortive attempt to improve the output stages?
No? Well just go back to about the middle of the last page :tongue:

N E way. I needed a break from all the other stuff I've been doing these last few months, so got the abandoned PCB and drawings out again, and started experimenting. First confirming instant and massive HF instability. Of all the things I tried last time, the one I didn't try was to find out what the open loop gain was (that's with zero feedback), so I isolated individual stages and measured it, then multiplied them all together. The result was a totally insane value!

The output 4 transistors had a gain of 10 so I reduced that to 3. That made quite a difference with a full load connected, but still some 'grass' with certain waveforms, and quite unstable when lightly loaded.

Then instead of a fully complementary driver stage I reverted to single transistor against a resistor, at the same time fitting a small bypass cap to roll of extreme HF.

This looks like a result... hopefully! I'm not seeing any problems at any signal level, and with no load, 15 ohms or 8 ohms loading.

DC stability looks good, and driving up to limiting gives symmetrical clipping with less than 5mV DC shift.

These tests are on a controlled PSU that provides a maximum of +/- 31V, and at that I get 20V RMS into 8 ohm. i.e. 50W. The Amp PSU gives +/- 36V so I'd expect to be able to push this to at least 24V RMS which gives me well over the original 60W spec.

I'll put up a new drawing when I'm satisfied it's all tickety-boo. I'll also need to re-do the distortion tests.

Re: D.I.Y. amp

PostPosted: Fri Jun 12, 2020 4:12 pm
by Hugh Robjohns
Good progress! :clap:

Re: D.I.Y. amp

PostPosted: Fri Jun 12, 2020 4:40 pm
by ef37a
Will, have you got Duggy Self's book on power amplifiers?

Really worth reading (for one thing he is very much NOT a fan of output stages with gain over 0dB!)

The final designs, if you included all the protection devices needed for a production amp are quite complex but for careful folk like us the bare circuit is not that bad a build and distortion levels in the third decimal place at 10kHz are easily met and THD at 1kHz barely readable on the best AP kit. That is for his class B design "Blameless Amplifier" The 20W class A jobby is nigh on perfect!


Re: D.I.Y. amp

PostPosted: Fri Jun 12, 2020 6:11 pm
by Folderol
I think you're missing the point a bit Dave. I quite accept there are many people who are far more knowledgeable about audio amplifier design than me, Mr Self being one of the best. However, if I used his designs instead of my own ideas it wouldn't be my amplifier, it would be my build of Mr Self's. Not the same thing at all.
A vanity project you might say :tongue:

Re: D.I.Y. amp

PostPosted: Fri Jun 12, 2020 10:17 pm
by ef37a
Folderol wrote:I think you're missing the point a bit Dave. I quite accept there are many people who are far more knowledgeable about audio amplifier design than me, Mr Self being one of the best. However, if I used his designs instead of my own ideas it wouldn't be my amplifier, it would be my build of Mr Self's. Not the same thing at all.
A vanity project you might say :tongue:

Ah. Quite right too. I thought you was on a mission for the ultimate amplifier.

SAY n'more!


Re: D.I.Y. amp

PostPosted: Sat Jun 13, 2020 7:39 pm
by Folderol

I'm getting more and more confident of the new design. Over the last few days I've tested it with a pair of highly inductive nominally 8 ohm dummy speakers, both parallel, one by itself, and both series, giving 4, 8, 16 ohm load doing static tests, and also pulsing the input with sine and square waves at low level and full power - no issues at all. I've also run it with nothing but a 2u2 mylar cap in series with a 0.47 ohm resistor, which must be far more reactive than any real load it will ever see!
So, drawings:

This is the test rig, and here is the failed one. As a reminder I include the current installed version that has been running for a few years now, but does have some DC drift issues, and asymmetric overload behaviour.

I can't do a distortion test yet. My old test kit ran on 2x pp3, and I forgot about it and left them in. They corroded and rotted the connectors. Fortunately they were sat in the bottom of the unit so nothing else was damaged. I've ordered the bits for a mains PSU to go inside it now. However, I'll make the rails +/- 15V to give me a bit more headroom.

Re: D.I.Y. amp

PostPosted: Sun Jun 14, 2020 11:21 pm
by Martin Walker
Folderol wrote:I've also run it with nothing but a 2u2 mylar cap in series with a 0.47 ohm resistor, which must be far more reactive than any real load it will ever see!

Ooh - that's a torture test and a half Will :shocked: :crazy:

Progress indeed if it's still stable into that combo :headbang:


Re: D.I.Y. amp

PostPosted: Sun Jun 21, 2020 5:47 pm
by Folderol
I was wrong about the distortion kit being OK :(
I can only guess that at some point it must have been sitting upside down with the battery gunk dripping on it, as one of the switches was so badly corroded bits fell off it when I tried to change its setting. However, I had a bit of luck in that there was a switch in one of my spares boxes that could be wired to suit.

So, with that sorted and the unit put together again, I started making tentative checks on the test rig amp. Initial results are looking promising, but I can't run at full power with that tiny heatsink for more than a few seconds, and that's not long enough to adjust the the distortion kit controls, then take readings. However at 1W out the distortion products are buried in the hum and noise. At 20W (the highest I dare go for this) I have a bit of a problem as the 'scope is picking up hum from somewhere, and I can't work out where. The good news is that at this power level and with a 1kHz input, it seems the total distortion is something like 75-80dB down.

Re: D.I.Y. amp

PostPosted: Mon Jun 22, 2020 12:14 pm
by ef37a
Will, have you tried recording the output, you only need a 5 sec burst, then putting the signal through Right Mark Analyser?

I have not gotten around to using it for THD measurement, just the RTA but it is all there.