Heathkit S-99 distortion after operation

[snowman_al]

Which inputs are you using? And what source?
PU1 and 2 and Aux use V1, 2 and 3.
Tape and Radio miss out V1 and feed to V2.

[qualityten]

I’m using Tape and Radio. I know that the Aux is much too sensitive, and I’ve not had satisfactory results with either PU 1 or 2 with vinyl.

I’m using a Philips CD624 player as music source and a Levell RC oscillator TG200 for square and sine waves.

[snowman_al]

I wonder if the output of the CDP is still overloading the input?
Have you tried a mobile phone as a source? You can turn it down.
Otherwise I am out of simple ideas.
You will have to look at feeding a signal into each next stage of the pre-amp and see where it stops distorting.
Not easy on the S-99 due to the complexity of the filter and tone networks and feed back from the anodes of the 3 triodes to the previous stage.

[qualityten]

Originally Posted by PJL

Where are we at? I am a bit concerned about the subjective test of distortion as it could be anything from the source to the speakers (and hearing!).

Are the output cathode voltages OK? v4(7)=V5(7)=10V
Is the concertina OK? voltages across R38=voltage across R40?

NOTE: The V5(1) grid voltage of 28V on the schematic is plainly wrong and the loading of your meter will further lower the measured value.

I've tried the amplifier with two different CD players and sets of speakers.

The output cathode voltages are okay, on or near 10V.

Thanks for confirming that the V5 grid voltage on the schematic is an error!

Yes, the concertina voltages are OK. Same drop across R38 and R40 (and other channel).

Thanks for all the diagnostic questions. The only components I wonder about are the 47pF ceramic capacitors (C17 and the other channel). When I replaced nearly all the components in the preamp, I used tiny ceramic capacitors there. Although the same value as before, I wonder if my replacements are not suitable? I don't understand their function between the bass and treble controls.

[qualityten]

As a p.s. I tried using my phone for a signal, but reducing the input doesn't really make any difference. Yesterday, as a hearing and sanity check, I tested my CD source and speakers with a different set of amplifiers of the same vintage (two Verdik Qualitytens--the Mullard 5-10 circuit--via just a pot) and there was no audible distortion.

[snowman_al]

Umm going to be something odd.
What voltage ceramic caps did you use? The ideally need to be 630 or more volts. You could easily prove if they are a problem, unsolder one end on each and try it with them disconnected, but they are in the power amp....

[qualityten]

I'm referring to the 47pF capacitors physically on the preamp board. They are part of the tone control circuit, marked C17 (and other channel) on the schematic, and I've just checked that they don't have any DC on them.

But, stepping back, replacing the balance control has made a difference and I think I need to try a range of recordings over a longer period. (I have just tried a different CD that I know very well and, after a short time of operation, can't hear any distortion.) Noting PJL's point about subjectivity, I'll give the amp a bit of soak testing with a variety of musical inputs.

So, I'll pause this thread, with thanks to all who have helped, especially snowman_al. it has helped me look harder at the circuit.

While the layout and the PCBs are awkward, it is completely hum-free. Even with the screen behind the preamp board removed.

David

[majoconz]

Copied from my post about the "Dansette smoking..."

I'm wary of ECL86's. An auction buy of a Blaupunkt "Sultan" sounded pretty good after a bulk replacement of the large C's and 'that' capacitor along with a largish resistor on the output transformer tags, a 1kohm 3 watt job. After 15 minutes or so it started to sound very 'bassy' which got worse and a hot smell - the 1kohm looked a bit toasty. All of the current through the 1k goes into the pentode side of the output ECL86. I attached a voltmeter across the 180ohm cathode resistor to monitor the current. Immediately after warm-up it was passing about 15mA and it stayed there for about ten minutes but then started to climb - when it got to 45mA panic set in. I asked a couple of friends at work and they both said 'that capacitor' which I had already changed but another one made no difference but then someone said 'grid emission' which I had to look up. Apparently ECL86's are renowned for it - I sorted though the spares and found four of them - two Mullards, one unreadable and a Telefunken - only one passed the test and stayed stable at 15 to 18mA - the Telefunken. I wonder if it has a slightly different construction but I couldn't see as it was well 'gettered'. So beware the ECL86! I'm sure that not the same for all circuits, perhaps it's just running near its limits in this one. The data says Va 250v and Ia 36mA so it's running about half the current.

[snowman_al]

Originally Posted by qualityten

I'm referring to the 47pF capacitors physically on the preamp board. They are part of the tone control circuit, marked C17 (and other channel) on the schematic, and I've just checked that they don't have any DC on them.

Apologies, I missed ''C17'' on the original post. I assumed it was the additional 47pFs in the power amp... Alan

[RichardGM]

Originally Posted by PJL

Where are we at? I am a bit concerned about the subjective test of distortion as it could be anything from the source to the speakers (and hearing!).

Are the output cathode voltages OK? v4(7)=V5(7)=10V
Is the concertina OK? voltages across R38=voltage across R40?

NOTE: The V5(1) grid voltage of 28V on the schematic is plainly wrong and the loading of your meter will further lower the measured value.

I did a double take when I saw the "concertina" reference. I do know what it means in this context, but two minutes earlier I had been on https://www.concertina.net/forums/ (as I am a concertina player).

[wd40addict]

I'm afraid ECL86s do this, I had to change them in my S-33H for this reason. Put the rogue on my VCM and after a while the anode current would shoot skywards. In fairness I've also seen EL33s and UL84s do this as well.

There's a lot packed into an ECL86, Mullard struggled and no-one makes a modern version. I fitted replacements and improved the ventilation around my amp, and so far so good. However I do wonder if Rk of 100R in the S-33H is pushing it a bit...

[qualityten]

I'm not sure I understand this grid emission issue with ECL86s, but I have checked the cathode voltages, and they don't drift upward over time. But, exaggerated 'bassiness' is one of the features of ECL86s I note in my Armstrong 227 receiver.

One possible clue to the location of my distortion is that at some point in trying different ECC83s in V3ab position I put in an ECC88 by mistake. The markings were a partially rubbed out, but I noticed that the heaters didn't light up. Strangely it did work anyway and without obvious distortion. Is this a clue?

[snowman_al]

It works without V3a/b because there is a direct connection from the bottom of the 10k resistor R27, in the tone control circuit, to the Normal / Reverse switch and so to the balance / volume control. (This is a feedback / forward circuit when the valve is operational.)

[qualityten]

Thank you Alan. Does this suggest that I should check the components in that feedback/forward circuit? What is the function of that circuit? Could I bypass it?

[snowman_al]

It is supposed to reduce distortion... And will reduce the gain of the V3 stage a little.

It is a piece of wire, you can only try removing it. Lift one end or other at R27 (and the other channel) and the Norm/Rev switch, which ever is easier.

I assume you have 1.3 volts on the cathodes, pins 3 and 8 of V3 and 150 volts on the anodes, pins 1 and 6.

[qualityten]

I lifted the connections at R27 (and other channel) and the result was obvious distortion in both channels. (Lifting it a the Norm/Rev switch breaks the signal completely.)

Restoring all connections, I listened once again to it without V3. The signal is clearer and the distortion seems to be in L hand channel. I'll do a bit of signal tracing.

[Valvepower]

Hello Alan,

If it helps I've found the balance pot I salvaged from the basket case S-99.

If it's of use it's yours FOC.

It's un-tested, but I recon it could scrub-up OK with some DeoxIT

Terry.

Originally Posted by snowman_al

At least try another ECC83 V3a/b. Lets eliminate it.

Reverse log or anti log pots are scarce. Usual dodge is to make a 'fake anti log pot' from a linear pot.
Find a 500kΩ dual linear pot that fits the S-99 hole and knob.
Use a 47kΩ (or 68kΩ) resistor to connect the centre tag (out to vol control) to the input tag (from the Normal / Reverse switch). One for each half of the pot.

It will work better than what you have (had). Have a look at https://sound-au.com/project01.htm#s5 Better Balance Control fig 8.

[dave cox]

Originally Posted by qualityten

Square wave testing on all the audio frequencies shows a clear trace, but this gets blurry at the edges as distortion sets in.

This may be significant, can you post a photo ?
Could be several things ...
(1) Oscillation on part of the waveform, usually at a much higher frequency. Will probably need a scope with a dual timebase to see this properly. Alternatively, lash-up a high-pass filter to remove the stimulus frequency.
(2) 50Hz pick-up, test with a 50Hz square wave and switch the scope between normal and line triggering.

dc

[qualityten]

Thanks for these recent comments. I've waited till today to replace the tiny 47pF ceramic caps with better silver mica ones. Subjectively, I think that has made an improvement.

I've also been wondering why the schematic wrongly shows a voltage of 28VDC on the triode grid of V5a. I replaced the 0.05uF coupling capacitors but still get 14V here. Is this in the right ballpark?

Dave, thank you for the advice on where to look. I don't have a dual timebase scope, but I'll take a look at what the square wave looks like now and post a pic, if significant.

Terry, thanks for the offer of a replacement balance pot. I'll PM you my address.

David

[PJL]

Originally Posted by qualityten

I've also been wondering why the schematic wrongly shows a voltage of 28VDC on the triode grid of V5a. I replaced the 0.05uF coupling capacitors but still get 14V here. Is this in the right ballpark?
David

Firstly, the grid is anchored to a voltage divider on the cathode giving V=28*47000/(47000+4700) = 25.45V. Then there is a further voltage division caused by your meter and the 470K grid resistor but I am not sure what meter you are using? If it is a 10M input resistance DMM then I would say it is low and worth investigating BUT earlier you said both channels showed the same result suggesting it is probably not a fault.

Fast rise/fall times of square wave go too fast for the delay in the negative feedback to compensate and typically result in some ringing. As the ringing is dependent on the loop delay, it is affected by the load impedance on the amplifier.