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 21st Mar 2023, 10:35 pm   #1
cathoderay57
Nonode
 
Join Date: Oct 2017
Location: Bristol, UK.
Posts: 2,364
Default LF Instability TAD EL34B-STR

Good evening all. I have a home built push-pull stereo power amplifier using 2 x EL34 per channel. It was a project published in Elektor Electronics magazine in April 2003. While it is (was) working fine I happened upon a pair of TAD EL34B-STR valves that I wanted to try. Currently I have an odd selection of valves in one channel: here's how the existing 2 valves check out on the VCM 163:

CVC Chelmer EL34: Ia=79mA (75) gm=11mA/v (11)
Mullard EL34: Ia=83mA (75) gm=11mA/v (11) with 25M h/c leakage.

The amp works absolutely fine with these because, although slightly unmatched, I can adjust the biassing using the P2 and P3 presets shown in the attached circuit extract to set each valve to exactly 50mA standing current i.e. 0.5v on each of the grid bias resistors R24/R25.

Here are the VCM 163 test results for the TAD valves:

TAD EL34B-STR #1: Ia=76mA (75) gm=10.5mA/v (11)
TAD EL34B-STR #2: Ia=71mA (75) gm=10.0mA/v (11)

So I plugged the TAD valves in, using a 10 Ohm resistor as speaker dummy load. I can achieve a steady bias at 30mA (0.3v on the cathode resistors) but once I advance the preset to increase the standing current to 40mA a slow oscillation in the valve standing current starts up and increases in amplitude swinging between about 30mA and 80mA on each valve, at 1Hz or maybe a little less. The swing in standing current is reflected in the other channel whose valves were unchanged. I have checked the HT voltage which is about 480v (as it has always been) and the two main HT smoothing capacitors (2 x 470uF 450v in series) measure 286uF using the Fluke 110 multimeter capacitance tester.

I tried using the TAD valves, one at a time, paired with the original CVC Chelmer valve, and got the same oscillation problem. Putting the Mullard and Chelmer valves back in, the set-up proceeded normally and was stable. Any clues as to why the TAD valves are playing up, and where should I start to look for circuit instability? Cheers, Jerry
Attached Files
File Type: pdf Amp extract.pdf (76.1 KB, 44 views)
cathoderay57 is offline  
Old 21st Mar 2023, 11:38 pm   #2
trobbins
Heptode
 
trobbins's Avatar
 
Join Date: Oct 2009
Location: Melbourne Australia
Posts: 898
Default Re: LF Instability TAD EL34B-STR

Can you post the complete circuit schematic - to show the global feedback loop - and a photo or two of your circuit layout ? What OPT are you using, and what was specified in the project parts list?

The TAD valves would change the forward loop path gain. Your original configuration may have been borderline stable.

The OPT likely has its P-P inductance acting as a significant double-pole roll-off down near 1Hz, and the change in dc bias current (even though close to net-zero balanced at the terminals) is moving that filter gain/phase around a bit and the closed loop is just reaching into an unstable condition.

I'm guessing you don't want to change OPT's, so the other options are to modify the low frequency roll-off circuits embedded in the forward path. Given you may be close to borderline stability for starters, then that may mean changing simple CR coupling filters to include a series CR across the C, to form a shelf network, which was a common method used from the late 1950's to improve LF stability margins. Assessing such LF gain and phase response is not simple, and typically requires an X-Y scope plot and a low freq sig gen, and the obtainable measurement results can be quite coarse (as any result is based on the X-Y circle-ellipse). Changing the LF stability can also change the closed loop low frequency corner and distortion performance as the feedback level is being modified down there.

Last edited by trobbins; 21st Mar 2023 at 11:45 pm.
trobbins is offline  
Old 21st Mar 2023, 11:44 pm   #3
Gabe001
Octode
 
Join Date: Oct 2019
Location: Stockport, Greater Manchester, UK.
Posts: 1,612
Default Re: LF Instability TAD EL34B-STR

What is your oscilloscope showing at the speaker dummy load end?
Gabe001 is offline  
Old 22nd Mar 2023, 8:38 am   #4
Michael Maurice
Moderator
 
Michael Maurice's Avatar
 
Join Date: Feb 2006
Location: Wembley, Middlesex
Posts: 7,219
Default Re: LF Instability TAD EL34B-STR

I would be contacting TAD and asking them why their valves are causing this problem.
__________________
Forum Moderator

http://www.michaelmauricerepairs.co.uk/
Michael Maurice is offline  
Old 22nd Mar 2023, 9:16 am   #5
cathoderay57
Nonode
 
Join Date: Oct 2017
Location: Bristol, UK.
Posts: 2,364
Default Re: LF Instability TAD EL34B-STR

Good morning, and thanks for your replies. I have asked Elektor Magazine for permission to post the circuit. This might take some time. The circuit is based on the Mullard 20W design published in the Green Book. The main differences are a triode-connected EF86 and a separate supply for the negative bias for the EL34s, also an AC balance adjustment for the output stage. There is a reduction in the amount of negative feedback by 20dB. I can say that the feedback network comprises a 680pF capacitor in parallel with a 3k3 resistor connected between the speaker secondary and the midpoint of a 390R and 100R resistor in the cathode circuit of the first valve, a triode-connected EF86. The second valve is an ECC83 phase splitter, cathode coupled. The OPT is a Lundahl LL1620/P-P. Primary/secondary turns ratio 4 x 19.2 / 8 x 1, primary self-inductance 300H, primary leakage inductance 13mH, primary impedance 6k, secondary 8R. I'll try to post an image of the output waveform later today. I went round the loop (forgive the pun) attempting a Bode plot of the phase/frequency response of my Quad IIs recently and didn't have a lot of success but I'm prepared to have another go with this amp as a last resort.
Kind regards, Jerry.
Attached Thumbnails
Click image for larger version

Name:	20230322_074725.jpg
Views:	63
Size:	53.5 KB
ID:	275527   Click image for larger version

Name:	20230322_074945.jpg
Views:	61
Size:	77.4 KB
ID:	275528  
cathoderay57 is offline  
Old 22nd Mar 2023, 9:34 am   #6
cathoderay57
Nonode
 
Join Date: Oct 2017
Location: Bristol, UK.
Posts: 2,364
Default Re: LF Instability TAD EL34B-STR

Wow, that was quick. I got an approval to post by Elektor by return email. circuit posted. Source: Elektor 4/2003 www.elektormagazine.com.
Attached Thumbnails
Click image for larger version

Name:	2023-03-22 08.34.29.jpg
Views:	91
Size:	69.6 KB
ID:	275531  
cathoderay57 is offline  
Old 22nd Mar 2023, 9:57 am   #7
cathoderay57
Nonode
 
Join Date: Oct 2017
Location: Bristol, UK.
Posts: 2,364
Default Re: LF Instability TAD EL34B-STR

Having re-read the construction article, I found this: "In the original design, the coupling capacitors for the output valves (C9 & C10) had a value of 470nF. The current through the output valves proved to have rather large fluctuations at a very low frequency (0.2-0.5 Hz), which were also present at the speaker output. This was probably due to small variations in the negative grid voltage. Since the fluctuations have a small amplitude and the output transformer has a large self-inductance, they are not blocked by the output transformer and they find their way to the amplifier input via the negative feedback network. This phenomenon was reduced to an acceptable level by decreasing the value of C9 and C10 to 100nF".
cathoderay57 is offline  
Old 22nd Mar 2023, 11:02 am   #8
Gabe001
Octode
 
Join Date: Oct 2019
Location: Stockport, Greater Manchester, UK.
Posts: 1,612
Default Re: LF Instability TAD EL34B-STR

The cure for LF instability is different to that of the HF instability you had in the quad, so a different approach is required

From Morgan Jones "building valve amplifiers"

"Because motorboating is a low-frequency (1 Hz) oscillation due to unwanted power supply coupling, experiment to see if increasing smoothing capacitance at one stage changes the frequency. If you can change something, then you must be near to the source of the fault. The best cure is to reduce the
HT source resistance. A regulator is ideal, but reducing an HT series resistor might work. Alternatively, reducing smoothing capacitance might be acceptable, depending on hum. If all else fails, you could resort to the traditional cure of reducing the value of audio coupling capacitors ......."

Electrolytics drying up may be compounding the problem by changing the LPF cutoff frequency

Hope this helps
Gabriel

Last edited by Gabe001; 22nd Mar 2023 at 11:12 am.
Gabe001 is offline  
Old 22nd Mar 2023, 12:10 pm   #9
trobbins
Heptode
 
trobbins's Avatar
 
Join Date: Oct 2009
Location: Melbourne Australia
Posts: 898
Default Re: LF Instability TAD EL34B-STR

The schematic appears to show 10uF rail filtering for the input stage and the PI stage, which are circa 16kohm shunt at 1Hz. The supply rail droppers are 10k and it looks like 100k to the input stage.

The 100k dropper and 10uF input stage filter loses its ability to decouple below 1Hz, with the input stage gain increasing as the 100k starts to become part of the EF86 triode anode load. In some amps like the Williamson that aspect was turned in to a LF stability tweak, but in your amp it would seem that the designer didn't make the LF stability margin enough, and extra filtering beyond 10uF would appear to be necessary.

Most modern amps would not purposefully use such a low filter capacitance value, but it may be more of a concern for your amp given a very high OPT primary inductance which is pushing its LF roll-off to below 1Hz (eg. the Williamson's large OPT had a roll off of a few Hz).
trobbins is offline  
Old 22nd Mar 2023, 5:39 pm   #10
cathoderay57
Nonode
 
Join Date: Oct 2017
Location: Bristol, UK.
Posts: 2,364
Default Re: LF Instability TAD EL34B-STR

Thanks again for the advice. I added a second 10uF cap in parallel with C8, the PI stage filter and it improved things a lot. There is still sporadic oscillation in output valve currents but it is reduced in amplitude compared to before, and also seems to come and go on one or the other channel. When oscillation is present, the speaker output on the scope is showing less than 1mV LF amplitude, barely above the noise. I've therefore ordered 4 x 22 uF 500v electrolytics to replace C3 (input stage) and C8 (PI stage) on both channels and we'll see how we go. I noticed that the person who built the amp used 10uF 450v caps for these and so bearing in mind the HT rail sits at 480v prior to V1 and V2 warming up then C8 will have had a hard life. C3 is probably OK as far as voltage rating is concerned since it is protected by the 200v zener D1, but I'll replace it with the higher capacitance value to help suppress the LF instability. I think the HT voltage is higher than depicted on the diagram probably because this project was designed for European 220v mains and I have between 235 and 240v. I'll post an update when the caps arrive. Cheers for now, Jerry.
cathoderay57 is offline  
Old 27th Mar 2023, 5:34 pm   #11
cathoderay57
Nonode
 
Join Date: Oct 2017
Location: Bristol, UK.
Posts: 2,364
Default Re: LF Instability TAD EL34B-STR

Hi folks, 22uF caps arrived and fitted. Unfortunately, it still oscillates with the TAD valves fitted, but maybe acceptably. In Post #7 the designer noted that he only managed to suppress the LF oscillation to "an acceptable level" so maybe it's now as good as it gets. The voltage swing on the TAD EL34 cathodes is now between 0.4-0.6v which equates to a combined anode/screen current swing between 40-60mA (desired setting 50mA) at about 1Hz. Jerry
cathoderay57 is offline  
Old 27th Mar 2023, 5:52 pm   #12
Radio Wrangler
Moderator
 
Radio Wrangler's Avatar
 
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 22,800
Default Re: LF Instability TAD EL34B-STR

LF instability is harder to avoid than HF!

You also get phase shifts building up at the LF end of an amplifier's range. If you try to roll off the feedback at the LF end so that it kills the loop gain before the main amplifier has rolled its own gain below the trouble threshold, you tend to get a whacking great bump in the LF frequency response. Oddly enough, you don't want coupling capacitors too large.

Trad valve architecture has compromises which paint you into two corners at once. I wimped out and opted for DC coupling and no output transformer in the end.

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 27th Mar 2023, 6:21 pm   #13
cathoderay57
Nonode
 
Join Date: Oct 2017
Location: Bristol, UK.
Posts: 2,364
Default Re: LF Instability TAD EL34B-STR

Thanks David. The designer tried to circumvent this problem by reducing the EL34 grid couplers from 470nF (500nF per original Mullard circuit) to 100nF. I wonder if it would be worth trying 47nF? I have seen circuits using couplers ranging from 500nF down to 22nF. Jerry
cathoderay57 is offline  
Old 27th Mar 2023, 11:48 pm   #14
trobbins
Heptode
 
trobbins's Avatar
 
Join Date: Oct 2009
Location: Melbourne Australia
Posts: 898
Default Re: LF Instability TAD EL34B-STR

Jerry, the designer used a 'step network' for HF stability on the input stage anode - its the series RC that is in parallel with the anode resistor for V1. The step network was reasonably well known in the 1950's as a way to improve stability margin by modifying the main path gain and phase in a frequency band between the nominal audio band, and the frequency region where zero-gain and zero-phase margins influence stability (eg. between say 10-20kHz and say >150kHz for hi-fi HF stability in amps like the Williamson and in your amp's design).

The same can be done for LF stability, ie. introduce a step network that imposes a gain and phase step between the lower audio range (eg. 30-60Hz) and the region where zero crossing is occuring (eg. somewhat at or below 1Hz in your amp). In the past, that has been done on the two 100nF coupling caps.

The link below discusses how this can be applied to the Williamson amp, and Section 8(c) on p.21 in the sub-section on 'Phase splitter stage output CR step/shelving network' goes to the particular details with example part values, and links are provided to how Partridge and GEC and Kiebert and Patrick Turner used the step network in their amps.

This change is based on basic RC network appreciation, but taken to the next level of electronics 101 by Roddam (ref #20) and Learned (ref #19), but still understandable by the masses in a WW article in 1951. It may be a bit daunting at first to work out component values for any modification, so the forum can help with that.

https://dalmura.com.au/static/Willia...ign%20info.pdf
trobbins is offline  
Old 28th Mar 2023, 6:13 am   #15
Gabe001
Octode
 
Join Date: Oct 2019
Location: Stockport, Greater Manchester, UK.
Posts: 1,612
Default Re: LF Instability TAD EL34B-STR

Quote:
Originally Posted by cathoderay57 View Post
Thanks David. The designer tried to circumvent this problem by reducing the EL34 grid couplers from 470nF (500nF per original Mullard circuit) to 100nF. I wonder if it would be worth trying 47nF? I have seen circuits using couplers ranging from 500nF down to 22nF. Jerry
If you make this too small, smaller than the subsequent poles set by the transformer and el34 cathode RC network, you decrease the low frequency range over which the feedback applies. This leads to a higher LF peak at the anode at the stage where feedback is applied, which will be shifted to the right I.e a few Hz higher than the low frequency value of the coupling cap you've changed. You may need to compensate for this with an RC filter at the input to avoid overloading the stage. As others have pointed out, it can be a pain to get right as one change may have an effect both upstream and downstream. Still, it may be worth a try.

When I built my stereo el34 amp for my sister in law I couldn't get the feedback compensation right and finally ended up using less feeback than I initially intended. LF overload was part of the issue and the circuit was becoming increasingly complicated. A good stable high feedback circuit is priceless.

The shared article proves that it can be done mathematically. I've saved it for a good read (thank you). Usually this very important element only gets a fleeting reference. Lt spice can help and it's quite good at frequency response modeling if you're adept at modeling OPTs on it (sadly beyond my current skill level).

Last edited by Gabe001; 28th Mar 2023 at 6:30 am.
Gabe001 is offline  
Old 28th Mar 2023, 6:45 am   #16
Radio Wrangler
Moderator
 
Radio Wrangler's Avatar
 
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 22,800
Default Re: LF Instability TAD EL34B-STR

In the article trobbins links, fig 2 shows the bass peaking in the closed loop bode plot quite nicely.

Without care it can be substantially larger, and if you're really unlucky you hit rumble frequencies or speaker resonances.

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 28th Mar 2023, 7:54 am   #17
trobbins
Heptode
 
trobbins's Avatar
 
Join Date: Oct 2009
Location: Melbourne Australia
Posts: 898
Default Re: LF Instability TAD EL34B-STR

The venerable Williamson amp had a plethora of different OPT's thrown at it over a few decades. Many OPT's didn't get to the 100H PP primary inductance target (especially when bias imbalance occurred), and suffered from a resonance that was at or above the coupling cap corner frequencies, which meant that all hell could break loose in the vicinity of 5-20Hz (depending on the OPT). Williamson managed to get his demo OPT to have a resonance just under the coupling cap corners, and had to add an uncommon circuit tweak relating to the 8uF C1 and C2 supply rail filter caps to just get his amp borderline stable.

From the start of the 1950's, those in the know and not in a penny-pinching bind could make or purchase a suitable OPT that bettered what Williamson could muster in the mid-1940's, but many amp manufacturers and DIY'ers were caught out, and that was just the LF stability issue....
trobbins is offline  
Old 28th Mar 2023, 8:25 am   #18
cathoderay57
Nonode
 
Join Date: Oct 2017
Location: Bristol, UK.
Posts: 2,364
Default Re: LF Instability TAD EL34B-STR

Thanks all. I'll have a read and experiment over the next few days. Jerry
cathoderay57 is offline  
Old 28th Mar 2023, 8:44 am   #19
peter_sol
Octode
 
Join Date: Aug 2003
Location: Welwyn Garden City, Herts. UK.
Posts: 1,906
Default Re: LF Instability TAD EL34B-STR

Bear in mind that the original Mullard circuit was for 20 watts using cathode bias.
The fixed bias versions are to force more than 20 watts from the valves. Of course quality, stability and everything else suffers.
The whole circuit must be based around the output transformer and feedbacks and coupling caps etc trialled to get stability. good luck with your experiments.
i suspect the new valves may have strange interelectrode capacitances.
peter_sol is offline  
Old 28th Mar 2023, 9:03 am   #20
Michael Maurice
Moderator
 
Michael Maurice's Avatar
 
Join Date: Feb 2006
Location: Wembley, Middlesex
Posts: 7,219
Default Re: LF Instability TAD EL34B-STR

Quote:
Originally Posted by peter_sol View Post
Bear in mind that the original Mullard circuit was for 20 watts using cathode bias.
The fixed bias versions are to force more than 20 watts from the valves. Of course quality, stability and everything else suffers.
The whole circuit must be based around the output transformer and feedbacks and coupling caps etc trialled to get stability. good luck with your experiments.
i suspect the new valves may have strange interelectrode capacitances.
Which is why I suggested you contact TAD.
__________________
Forum Moderator

http://www.michaelmauricerepairs.co.uk/
Michael Maurice is offline  
Closed Thread

Thread Tools



All times are GMT +1. The time now is 6:56 pm.


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.