UK Vintage Radio Repair and Restoration Powered By Google Custom Search Vintage Radio and TV Service Data

Go Back   UK Vintage Radio Repair and Restoration Discussion Forum > Specific Vintage Equipment > Vintage Audio (record players, hi-fi etc)

Notices

Vintage Audio (record players, hi-fi etc) Amplifiers, speakers, gramophones and other audio equipment.

Closed Thread
 
Thread Tools
Old 11th Oct 2019, 11:31 pm   #1
Michael Maurice
Moderator
 
Michael Maurice's Avatar
 
Join Date: Feb 2006
Location: Wembley, Middlesex
Posts: 7,219
Default SONY TA-FA3 HF noise

I am repairing this Sony amplifier, the circuit is below in both PDF and picture format.

The thing is that I found R625 to be rather cooked, so I replaced it but the new one also cooked. Further investigations showed a very high frequency on the output, either around 330KHz or 3.3MHz (1.5 divisions set to 2uS)

I tried putting some low value capacitors 15pF between the gate and the source of the output transistors but it made no difference.

The amplifier works but the heatsinks get very hot.

I find it interesting that the zobel networks for each channel (shown with the blue line) dont go to ground but are connected to each other. However in practice they dont!

I'm not happy returning it to the customer, the original fault was dry joints, but I've no idea how to rectify this.
Attached Thumbnails
Click image for larger version

Name:	TAFA3.jpg
Views:	112
Size:	100.0 KB
ID:	191726  
Attached Files
File Type: pdf TAFA3.pdf (223.3 KB, 70 views)
__________________
Forum Moderator

http://www.michaelmauricerepairs.co.uk/
Michael Maurice is offline  
Old 12th Oct 2019, 8:32 am   #2
Silicon
Nonode
 
Join Date: Jun 2007
Location: Coulsdon, London, UK.
Posts: 2,152
Default Re: SONY TA-FA3 HF noise

I noticed that the feedback wire from the power stage output to the preamplifier board goes through a linking cable to C511 and C611, etc.

This part of the circuit should be checked out for continuity, voltages, etc.

The oscillation could be in the preamplifier stage.
I would be tempted to 'ground' the signal going into the power amplifier but I am not clear yet how to go about it.
Silicon is offline  
Old 12th Oct 2019, 10:32 am   #3
samjmann
Heptode
 
samjmann's Avatar
 
Join Date: Feb 2014
Location: Nottingham, UK.
Posts: 645
Default Re: SONY TA-FA3 HF noise

C511/611 are part of the decoupling in the feedback loop. I've had problems with caps in this position before, but low value caused maximum negative feedback. If the basic DC voltages are correct then the chances are the transistors are all OK. The resistor is cooking no doubt as the series capacitors connected to it offer a low impedance path to the 3.3MHz HF. I've had some tone control networks do odd things, as in the above post, decoupling the input to the output stage, will show if the fault is early on.
samjmann is offline  
Old 14th Oct 2019, 5:43 pm   #4
dave cox
Nonode
 
dave cox's Avatar
 
Join Date: Jan 2009
Location: Bristol, UK.
Posts: 2,059
Default Re: SONY TA-FA3 HF noise

As well ...

Main 'miller integrator' cap is C607 - this one will see quite some voltage swing. C604 is going to slow things down as well. If either of these is out of spec it would likely cause oscillation.

C610, again, will get quite stressed - it is rolling off the closed loop gain.

A problem with electrolytics C611 (as mentioned), C606, C608 & C611 could cause issues as well.

Even the layout of the linking cable could come into play!

dc
dave cox is online now  
Old 14th Oct 2019, 7:41 pm   #5
Radio Wrangler
Moderator
 
Radio Wrangler's Avatar
 
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 22,798
Default Re: SONY TA-FA3 HF noise

I agree that the shunt Zobel network seems to be missing a ground

C607 creates the main pole in the amplifier's OPEN loop response
C604 creates another pole, with R605 adding a zero in the open loop response
C610/R610 although in the 'feedback' path creates a zero in the open loop path with R607 and C608 creating a pole.

Once at higher frequencies you get the unintentional poles which are implicit in devices and aren't to accurately defined.

The game is to use the resistors and capacitors to roll off the open loop gain down to unity gain or less before the total phase lag builds up to 180 degrees. Otherwise your negative feedback is phase shifted into positive feedback before the gain has gone below the self-sustaining level... and that means the whole loop will oscillate.

Of course the designer wants plenty of open loop gain in the audio range because that lowers distortion, lowers output impedance and lowers noise. But that makes the job harder rolling it all off before stray poles bring in more phase shift than you can handle.

A pole is like an RC lowpass section. Below its 3dB frequency its response is flat, above it the response falls at 20dB per decade. Phase lag is zero at low frequencies, 45 degrees at the 3dB frequency, and levels out gently at 90 degrees as frequency gets higher. A zero is the opposite sort of thing. Gain goes up at 20dB per decade, and the phase shift is a LEAD. still 45 degrees at the 3db frequency.

This is all quite complicated to calculate from the design point of view, but there are known solutions which you tend to start from and do a bit of fiddling for individual circuit designs.

In general, adding extra capacitors will add an extra pole and thus extra phase lag. Inevitably the lag gets effective before the roll off of the extra pole does anything much useful. Expect added capacitors to make things significantly WORSE. I know everyone's instinct is to roll off the gain more so it hasn't got gain enough to oscillate at that frequency any longer. But adding a pole makes it oscillate at a lower frequency and more strongly.

The design is a race, to get the amplitude to roll off quickly enough to reach the magic unity value, before the phase lag has built up to 180 degrees.

If such a circuit is unstable, then something has increased the gain so you have further to roll off, or else something has added lag somewhere around the loop.

So you're looking for something which creates either an increase in gain, or an increase in phase shift.

Just trawling through it, measuring all components might show something.

Ideally, the loop can be broken at signal frequencies, but with a very slow filter added to allow DC feedback to operate, and then the open loop gain and phase can be plotted starting with one stage and working onwards... comparing results with the design intent (Or you could get LTspice to calculate what you should see.)

A lot of work for a paid job to fix one amplifier. Needed when you're designing one and having trouble with the first prototype.

Lift R601 to check it isn't coming from an earlier stage!

Has anyone been here before you? Any changed transistors for example.

Sending it back out with an oscillation risks burning out tweeters.

At least it isn't an STK module!

David
__________________
Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done
Radio Wrangler is online now  
Old 14th Oct 2019, 8:32 pm   #6
John_BS
Octode
 
John_BS's Avatar
 
Join Date: Dec 2008
Location: Wincanton, Somerset, UK.
Posts: 1,756
Default Re: SONY TA-FA3 HF noise

Which, if any, components have you changed as part of the repair?

John
John_BS is offline  
Old 17th Oct 2019, 8:57 pm   #7
samjmann
Heptode
 
samjmann's Avatar
 
Join Date: Feb 2014
Location: Nottingham, UK.
Posts: 645
Default Re: SONY TA-FA3 HF noise

As ever RW sets the standards which I (and a few others I suspect) would love to aspire to!

The term 'pole' used in this way I've not heard before. does this mean the point of resonance for a LC circuit?

Looking at the job from a purely service engineer point of view, if only one channel oscillates then there is a fault rather than a design problem. This is assuming that the amp hasn't had some previous 'repair' etc.

I recently had a fault on a Dolby IC circuit on a Sansui cassette deck. All the input voltages were way off, in the end it turned out to be a 15nf polyester cap in a LPF circuit that was leaky. We've in the past assumed that these parts were almost always perfect. The capacitor measured perfect for leakage, but read 25nf on a capacitance meter! This part was around 40 years old.

SJM.
samjmann is offline  
Old 18th Oct 2019, 7:24 am   #8
Radio Wrangler
Moderator
 
Radio Wrangler's Avatar
 
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 22,798
Default Re: SONY TA-FA3 HF noise

Quote:
Originally Posted by samjmann View Post
The term 'pole' used in this way I've not heard before. does this mean the point of resonance for a LC circuit?
SJM.
Yes, and other things as well.

You get introduced to the things in any decent course on electronic design. But they teach it **** about face. The shove all the mathematical proofs at you before showing you what use they are (if they ever get that far) So you just let the maths whistle past your ears, paying no attention because it seems pointlessly abstract. If they then do show you what can be done with them, it's too late, you've missed the maths and you don't follow the language. The universities which understand this are very special indeed.

This is a pity because they are an absolutely essential tool for doing anything with feedback. Ever since they were devised, pole-zero techniques have been running a special offer: Learn about them and you also get given the keys to filter design, phase-locked loops and antenna arrays. For free.

Vandalism can be fun

On a map, you can fix your position by two numbers, one giving east-west axis, the other north-south.

You can move E-W quite independently of N-S.
The E-W axis is at right angles to N-S
They are said to be orthoganal.
Doing a 90 degree left turn will change you from facing East to facing North, or N to W or W to S or S to E.

"90 left!" is something mathematicians would call an operator. Outside maths it sounds more like something shouted by the navigator of a rally car.

Do this operator once and you start moving in a direction completely unrelated to your first direction.

Twice reverses your direction

Three times sends you the opposite way in that unrelated direction

Four times puts you back as you were.

This is a pattern.

Plenty of other operators exist, as well as the maths of more dimensions than two, but this is enough for our purposes here.

Going north has no real meaning if you are tied to the E-W axis. You could imagine northishness, though even if you couldn't do it.

Someone invented the idea of having 'complex' numbers, a pair of numbers. One saying how much real, the other how much imaginary value there is. It's like a navigational fix on a map, or a national grid reference. So you can plot these numbers on a graph, just like on a map. It's called an Argand diagram. Mathematiciand put a little flag, i, to mark the imaginary part.... like 6+i7. Electronics folks would get confused to hell using i for something other than current, so we use j instead (and this confuses mathematicians)

In electronics, these complex numbers are a handy way of taking into account phase shifts, which we need when inductors and capacitors get involved.

The reactance of a capacitor is 1/(2*pi*f*C)

But we could say the impedance of a capacitor is 1/(j*2*pi*f*C) and that j takes care of the 90 degree phase shift in capacitors. You can prove that 1/j = -j
So capacitors give negative imaginary impedance (true!)

The reactance of an inductor is 2*pi*f*L
so impedance is j*2*pi*f*L and the j nor being reciprocal gives us positive imaginary impedance.

So we can use all the familiar equations for combining impedances, analysing networks etc, and if we use the complex number forms of impedance, it will keep us honest and do all the phase work for us.

Remembering that doing the j operator twice reverses your direction j * j = -1
so j squared = -1 so people say that j is the square root of minus 1.... which everyone knows can't exist... but hey, it's imaginary, it's not real, so that's not only OK, it's necessary!

Now people find some things irresistable. Give 'em a button witha big sign "DO NOT PRESS" and you know what's going to happen

So there we were with these complex number thingies and we already had equations to describe all sorts of circuits and how they behaved over frequency... so someone wondered what if we shoved in a complex value for frequency?

"Don't be silly, it can have no real meaning" would have been the inevitable reply. So someone did it. and analysed the result.

All hell broke loose!

A simple R-C lowpass when analysed, came up as infinite gain at one imaginary frequency.

Gain out of a passive circuit?

Impossible? Yes, but only at an impossible frequency

Put in a little signal at that impossible frequency and you get infinite output... Um, that's infinite power, infinite energy and will destroy the universe.

Was the big bang started when some being invented an imaginary frequency signal generator?

You've got to admit this is the ultimate in vandalism

Flip it on its head and it's a good indication that imaginary frequencies are genuinely impossible, though they can be imagined.

There'll be someone in some government institution behind razor wire working on the imaginary frequency bomb. At least until the administrators mis-understood and thought it was the bomb which was imaginary and cut the funding.

Having decided that imaginary frequency is impossible, we can still analyse at it as well as real frequencies and combinations of both.

For all our circuitry turns into map of where it's gain hits infinity and where it hits zero. Once we have these points plotted, a few simple geometric rules allow us to interpolate and get the gain (and phase) at any frequency real, or imaginary or a complex combination.

It's a useful tool, even though it uses (hopefully) impossible concepts.

An R-C lowpass gives a pole on the imaginary axis, at negative imaginary frequency (numerically equal to its -3dB frequency)

An L-C resonator gives a pole up near the positive real axis, near to its resonant frequency, but offset a little to the left (-ve imaginary) the less the offset, the less the Q

So all the messing about with differential equations to analyse complicated circuits and filters turns out to be replaced with some simple geometry: Pythagoras and SOHCAHTOA.

How it works looks complicated but it really amounts to some fancy mental gymnastics in using ridiculous though simple concepts. The end result is easy to use.

David (streaming cold, can't sleep)
__________________
Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done

Last edited by Radio Wrangler; 18th Oct 2019 at 7:29 am.
Radio Wrangler is online now  
Old 18th Oct 2019, 10:51 am   #9
John_BS
Octode
 
John_BS's Avatar
 
Join Date: Dec 2008
Location: Wincanton, Somerset, UK.
Posts: 1,756
Default Re: SONY TA-FA3 HF noise

I don't know if Michael is still stuck with this (it's gone quiet), but assuming the original problem was dry joints and not component failure, I'd have a good look at the grounding of (especially) the PA class A blocks, where the feedback lands. The path is via CNP400 pins 5 & 7 and then CNP450 of the panel board. The power amplifiers seem to have unusually high gain (c 150?): normally you'd expect around 30 or 40. C510/610 are introducing 45 degrees phase lead in the feedback loop at about 400kHz, suggesting either that a) this is where the closed loop gain is approaching unity , or b) Sony just wanted to tame the HF gain somewhat.

So, I'd rigorously check all the grounding resistances of all boards through all connectors.

John
John_BS is offline  
Old 18th Oct 2019, 11:45 am   #10
Radio Wrangler
Moderator
 
Radio Wrangler's Avatar
 
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 22,798
Default Re: SONY TA-FA3 HF noise

There seem to no decouplers to ground from either the positive or negative supply rails of either channel.

As far as I can tell the only grounds appear to be the top of C600 and the bottom of C582. THese are just the feedback divider groundy ends and what looks like anti-rfi caps on the amplifier inputs.

Noting that the two channels seem to have independent grounds, I suspect this amplifier may be one of those where everything is taken back to come central star point. It's a good idea for low frequencies but runs into problems at high frequencies... or anything with gain still surviving at high frequencies.

There are some electrolytics around such as the ones on the bases ofthe two constant current source transistors per channel, and the base of the cascode transistor in the voltage amplifier stage.

Sorry about descriptions without component numbers, I can't read many of them.

Ah, and one across the vbe multiplier.

I think the first step is to look at the workmanship for signs of anything being changed before. and a check for iffy soldered joints. THe transistors in that class A driver may have cooked their solder.

Next is to look at the integrity of grounding wherever the star may be.

Then a check of electrolytic ESRs.

THis is a somewhat weird amplifier, and its stability might not have been that good to start with. THere are things done here which seem to have been chosen to create advertising claims.

Note the 47 and 100 ohm resistors in the gates of the power fets. These are RF stoppers intended to stop the transistors going into an individual device oscillatory mode.

I'm really worried about that zoben shunt going between channels and its series capacitors.

What did they think they were doing?


David
__________________
Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done
Radio Wrangler is online now  
Old 18th Oct 2019, 12:35 pm   #11
John_BS
Octode
 
John_BS's Avatar
 
Join Date: Dec 2008
Location: Wincanton, Somerset, UK.
Posts: 1,756
Default Re: SONY TA-FA3 HF noise

Re Zobels: he wrote:

Quote:
I find it interesting that the zobel networks for each channel (shown with the blue line) dont go to ground but are connected to each other. However in practice they dont!
Might suggest they do go to ground and the schematic is wrong?
John_BS is offline  
Old 18th Oct 2019, 1:19 pm   #12
Radio Wrangler
Moderator
 
Radio Wrangler's Avatar
 
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 22,798
Default Re: SONY TA-FA3 HF noise

It would also solve the issue of two diodes all with their cathodes going to a place with no DC path.

So what other errors are there? In hifi gear deliberate errors to protect designs from copying have been known, but this looks like a genuine error.

David
__________________
Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done
Radio Wrangler is online now  
Old 18th Oct 2019, 1:30 pm   #13
ms660
Dekatron
 
ms660's Avatar
 
Join Date: Apr 2011
Location: Cornwall, UK.
Posts: 13,454
Default Re: SONY TA-FA3 HF noise

If the Zobel network isn't connected across the output it won't be doing its job.

According to what I'm looking at the Zobel network is transposed on the board layout when compared to the schematic, the resistor in the network as shown on the board layout is shown connected to 0v rail and the "top" capacitor is connected to the output.

On the schematic the connection of the Zobel network to the 0v rail is not shown.

So far as I can make out.

Lawrence.
ms660 is offline  
Old 20th Oct 2019, 6:09 pm   #14
samjmann
Heptode
 
samjmann's Avatar
 
Join Date: Feb 2014
Location: Nottingham, UK.
Posts: 645
Default Re: SONY TA-FA3 HF noise

Thanks RW for the reply to my question. Back in 1975, when I was on the second year of C&G 272, we were taught 'j'. It all passed over me, it seemed so esoteric, and why does this have any relevance to servicing? That said, circuits going into oscillation can be very difficult to sort out when repairing. I had a sync-separator transistor once cause UHF radiation, simply because someone had fitted the wrong type transistor...
SJM
samjmann is offline  
Old 20th Oct 2019, 8:08 pm   #15
Radio Wrangler
Moderator
 
Radio Wrangler's Avatar
 
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 22,798
Default Re: SONY TA-FA3 HF noise

So if we look at Michael's job, the unstable amplifier, there are three possibilities:

1) Something has happened, maybe a device or passive component changed which has created either excess loop gain, or excess loop phase lag and the intendd feedback loop is oscillating.

2) Something has changed to create an unintentional feedback path, which creates an extra loop, and this loop happens to be unstable (because it hasn't been intentionally designed) This could be a missing or failed decoupling capacitor allowing signal to loop back via a supply rail as one example.

3) Internal capacitances in transistors form feedback paths effective at RF frequencies and if the circuit presents low impedances to some of the transistor's connections, it can oscillate without needing anything else. The main feedback loop is not involved. For example having a filter capacitor to ground may look almost an RF short at high frequencies, but the path leaves a small amount of inductance. On a three terminal transistor of any flavour look for RF paths to ground (or other common reference) Having one is OK. Having two is risky. Three is asking for it. This is the mode that the addition of 'stopper' resistors in base or gate leads is normally used to combat. Ferrite beads on bases etc are used to give the effect of resistance at RF, but without introducing any change at DC.

David
__________________
Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done
Radio Wrangler is online now  
Old 23rd Oct 2019, 11:45 am   #16
Michael Maurice
Moderator
 
Michael Maurice's Avatar
 
Join Date: Feb 2006
Location: Wembley, Middlesex
Posts: 7,219
Default Re: SONY TA-FA3 HF noise

Quote:
Originally Posted by John_BS View Post
I don't know if Michael is still stuck with this (it's gone quiet),
John
Michael is on holiday in Israel enjoying the sun, and completely forgot about this amplifier! I will attend to it when I get back.
__________________
Forum Moderator

http://www.michaelmauricerepairs.co.uk/
Michael Maurice is offline  
Old 23rd Oct 2019, 12:14 pm   #17
Radio Wrangler
Moderator
 
Radio Wrangler's Avatar
 
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 22,798
Default Re: SONY TA-FA3 HF noise

It looks like a fair puzzle, so we're all waiting to see how it goes.

It'll wait. Enjoy the sun!

David
__________________
Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done
Radio Wrangler is online now  
Old 7th Apr 2020, 10:59 pm   #18
Michael Maurice
Moderator
 
Michael Maurice's Avatar
 
Join Date: Feb 2006
Location: Wembley, Middlesex
Posts: 7,219
Default Re: SONY TA-FA3 HF noise

Well the holiday has passed quite some time ago and no self isolating, i'm starting to look at those 'difficult' repairs!

To add to the problems, one channel's output is a -48V. The transistors appeared at first to measure ok, but one in particular Q512 was very suspect.

I disconnected the output transistors but the output was still at -48V. Only removing Q512 brought the output up to +48V which one would expect.

Q512 measured leaky between source and drain.

Now both the 2SJ313 and its compliment 2SK2013 are obsolete.

Cricklewood Electronics sell 2SJ449 and 2SK2161 or 2SK1350.

Most Mosfets transistors of this type are used in switch mode power supplies.

I could get a pair of 2SJ303/2SK2013 but they are very expensive.

I take it I'd be better fitting originals.

Any suggestions?
__________________
Forum Moderator

http://www.michaelmauricerepairs.co.uk/
Michael Maurice is offline  
Old 8th Apr 2020, 12:17 am   #19
Radio Wrangler
Moderator
 
Radio Wrangler's Avatar
 
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 22,798
Default Re: SONY TA-FA3 HF noise

Originals may be best, but you don't want to blow them up immediately.

This situation is one where having a pair of current limited bench power supplies would be good, so they could supply the positive and negative rails and you wouldn't need to power up the rest of the amplifier.

Thinking back to the oscillation, there is a small value capacitor in a miller integrator feedback position around that PNP cascode voltage amplifier stage on the driver board. This is the main compensation capacitor and is intended to dominate the roll-off. And a day or so ago, I think it was Ted spoke of other amplifirs where all the colpensation capacitors of a load of 'opamps' had gone open circuit (some sort of corrosion?) Could this have happened here? is the little compensation capacitor OK? It may look tiddly small, but it is magnified by the gain of the stage it is wrapped around, and its operation is crucial to stability.

David
__________________
Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done
Radio Wrangler is online now  
Closed Thread

Thread Tools



All times are GMT +1. The time now is 9:45 am.


All information and advice on this forum is subject to the WARNING AND DISCLAIMER located at https://www.vintage-radio.net/rules.html.
Failure to heed this warning may result in death or serious injury to yourself and/or others.


Powered by vBulletin®
Copyright ©2000 - 2024, vBulletin Solutions, Inc.
Copyright ©2002 - 2023, Paul Stenning.