Advance LF Generator Type SG66

[davidgem1406]

I have obtained this Advance signal generator in some sort of working order but does need quite a bit of work done.

The main amplifier uses 2 EL86 valves along with 2 stabilizers 0B2 / 0B2WA.

1 stabilizer valve has a white band at the top inside the glass. Normally this would be associated with air inside the envelope, so I bought a replacement.

Today I tested the valves used in the main amplifier (AVO 2 part tester) however the EL86's I found I had a problem testing. As the valve warmed up it would just run away, as fast as I turned the zero pot down it continued to run upwards until I had no more pot left I could just about bring it to zero. But I was unable to get any sensible reading from either valve.
Somehow I don't think either valve is faulty is there somebody that could check these 2 valves for me? (Not on an AVO 2 part)
I will probably obtain 2 replacements anyhow, just in case.

I thought I would try to test the working of the stabilizers and set the tester to the required anode and cathode settings (Not listed under test data).
I started at 80V on the anode and turned the meter to zero then switched to 100V and the first 0A2 struck, the replacement I bought 0A2WA struck at 80V. Even the one with the white inside the glass worked. I was quite surprised I was able to do that.

Dave.

[davidgem1406]

Can somebody test these 2 EL86's for me, my AVO 2 part doesn't seem to like them.
Thanks

Dave

[trh01uk]

Dave,

I have no knowledge of this sig gen at all, however the problem you describe sounds to me like a leaky capacitor (or several leaky capacitors). The fact that it "runs away" at a rate you can follow with the pot, suggests that the problem is pretty slow acting, and this fits with a slightly leaky capacitor.

Look for capacitors (particularly paper) that are isolating HT from the control grid of the valves. Those will probably be the culprit.

Richard

[WME_bill]

Advance SG66 and testing EL86 on Avo 2 Panel tester.
I am not sure whether you are having problems with the SG66 or the Avo. I seem to remember that the Avo 2Panel finds difficulty testing high Gm valves. The EL86 is a Low Impedance, high Gm (13ma/v) valve, intended for series connection in transformerless audio output stages, all developed long after the Avo was produced, so I am not too surprised. I can test your EL86, using a CT160, and compare with others I have. Send me a PM is you still want this done.
Very odd result with OA2 striking at 80V. Should be 185V or so, at DC. Perhaps that test is at AC. I hope the Avo Valve Tester experts will see this and advise.
If the Generator gives a reasonable sine wave output, I should think the valves are good enough, so as Richard "trho1uk" says, your trouble may well be capacitors.
Looking at the SG66. the squaring circuit V3+V4A is fed from the oscillator out, so if that seems alright, it suggests the Sine output stage V4B + V5,V6 is giving problems.
The two Stabilisers regulate the voltage to the oscillator stages to 198V. If the HT rail is significantly different, then your white stabiliser valve may indeed be duff. I am interested that you have 2 of OA2 (150C4), which would put the rail at 300V. The circuit shows a 90C1 (90v) + 108C1/OB2 (108v) = 198V. Perhaps the originals failed, and something was plugged in. Though it gives a fairly major change, which could be part of your odd results. What are the HT rail voltages, each side of the stabiliser dropper resistor R29, and at the anode of the EL86 - V5.
Let us know what you find. Bill m0wpn

[Brian R Pateman]

Bill is right about the EL86 on a two panel AVO. It was never designed to test valves of that type.

If this thread is really about testing EL86 on a two part tester then;

A) It should be re-titled (let me know and I'll do it)'

B) You have your answer - it can't be done, or not to provide any meaningful results.

[davidgem1406]

Hi all, thanks for the replies.

To clarify the valves 2 x EL86 were being tested in the AVO 2 part not running in the Advance generator, so capacitors are not in the equation.
However it transpires these can not be tested on this tester.

I made a typing error when I said 0A2, it should read 0B2 & 0B2WA, sorry for that. And yes the strike voltage was probably more AC than DC. You would think the Anode V on the AVO should be rectified but not so!

Thanks to Bill m0wpn for his offer to test the EL86's I will be in touch about that.

I now have replacement EL66's and 0B2WA's 2 off each. I have not tried them out as yet as I stripped the unit right down for cleaning. it was very, very dirty and some parts needed some repairs done such as cracked ceramic wafers on the frequency selector switch. I am now building it back up and have the main amplifier board to fit and wire that will complete the rebuild up to a workable point ready for some testing.

Dave

[ms660]

A lot of the earlier type of valve testers were DC on the anode but totally unsmoothed, normall full ripple full wave, normally a derivitive of the hickok patent.

Lawrence.

[davidgem1406]

Well at last I have rebuilt the generator and it is working including the original EL86's. I still need to check how well it is working and its accuracy.

I have also to repaint a strip / name plate that fits along the bottom front edge and to repaint the cabinet, but that's now a summer time job.

Unfortunately one of the original knob caps is missing.

It was a total strip and rebuild to be able to clean out the years of muck that had built up. The output board was stripped of all components for cleaning and rebuilt.

A couple of before and after pictures:

Best regards
Dave

[davidgem1406]

I have now picked up this one again after some time away from it.

So far I have checked the frequency of all ranges and run into a problem or two.

First the frequency is quite some way of on all ranges, as much as 20KHz.
Stability is awful as the frequency shifts quite rapidly up and down.
I was using another advance generator (J2E) connected to the Y axis and monitored by a frequency counter, whilst the SG66 was connected to the X axis.
I could then see quite clearly what was happening, whilst the J2E was rock solid the SG66 was drifting all over the place.

I have now run into the problem that the circuit of the generator does not tie up to the service info I have.
The 2 valves involved are V1 & V2. The service info states V1 is an 12AT7 but fitted is an ECC88, V2 is correct being an ECF80. The V1 base is definitely wired for an ECC88 as the heaters are on pins 4 & 5 and no center tap at pin 9. Pin 9 is the screen for the ECC88 and is indeed grounded.

The circuit wiring is not as the circuit diagram I have. For instance the selector switch (S1) wafer B wiper is shown to be connected to V1a grid.
In actual fact it is connected to V2 triode grid, from there it just gets worse.

I have considered the circuit diagram to be wrongly marked, such as V1a and V2a transposed, or even V1a and V1b transposed, but that is not the case as neither way ties in to the circuit diagram.

I must now ask does anybody know of an earlier or latter version of this generator where it uses an ECC88 and an ECF82 as the sine wave generators.

[Freya]

The avo manual says
In isolated cases,when testing certain types of pentode valves the initial anode current does not attain a definite maximum but rises gradually, making it difficult to read mav figures. A slight reduction of the screen volts will cure this trouble, without causing any appreciable errors.

works on my two panel getting 9.5 maV

edit: i tried same valves on the taylor 45c with switches set at 564 and they wont test, the needle vibrates near zero.

stephen

[XTC]

Originally Posted by davidgem1406

I have considered the circuit diagram to be wrongly marked, such as V1a and V2a transposed, or even V1a and V1b transposed, but that is not the case as neither way ties in to the circuit diagram.

I must now ask does anybody know of an earlier or latter version of this generator where it uses an ECC88 and an ECF82 as the sine wave generators.

I can't help with this specific item.

It's my observation that Advance circuits stayed very much the same for years and they usually added suffixes for changes in the cabinet style. I wouldn't expect a major circuit change and Advance not indicate that with the model number in any way.

If the circuit worked well with a 12AT7, an easily obtainable and reliable valve they even used in their rf sig gens, I think it's unlikely they would move to an ECC88 for the fun of it.

It's my guess that this sig gen has been experimented with and codged about, although if you look carefully, it's usually obvious where that has been done.

I'd be inclined to return it to the 12AT7 circuit, unless it does turn out that there was an Advance ECC88 variation.

Pete.

[davidgem1406]

Hello Pete,

Thanks for the input, however that thought had occurred to me, has somebody changed it all? I can see no signs of it having been changed it is all to good and looks original. I suspect this is a production change made for good reason, whether the change was from or to the ECC88, with the associated wiring changes, is another question.
I to have thought about changing it back and rewiring to use the 12AT7 again but am reluctant to do so. I would also have to check all the resistor & capacitor values used for frequency selection.

Dave.

[XTC]

I just find it odd that they'd make a substantial circuit modification and not change the name of the sig gen to SG66A or SG69 or whatever.

There are a few people offering SG66 manuals on the WWW. No A/B/C variants I could see with a quick look. It might be worth emailing them and asking if their version has an ECC88.

After a long period of not being used, some equipment takes a very long time to stabilise. You can also have problems with old carbon resistors doing strange things as they warm up.

From what I can see from your photos, this Wien Bridge oscillator has a variable capacitor, rather than a double pot, and corresponding high values of resistance in the Wien network. I think that high value carbon resistors tend to have more problems than low value ones.

My experience of Advance equipment is with SG62 and H-1 RF sig gens, and a J-1 AF sig gen which all worked to spec after an overhaul in which hardly anything had to be replaced. Then there's a VM77C millivoltmeter, which is just about OK after having had just about every component replaced. Not one of my favourite bits of kit, but a lot of people have them and think they are fine.

Pete.

[davidgem1406]

Hi Pete,

I have looked further into the other parts of the circuit and the changes extend much further. Undoubtedly there are different version of this generator, but nothing mentioned on the unit itself not even on the model/serial label.

The range selection components are wired differently also the amplifier stage has changes that include the omission of some components, RV15, R66 and R27, as a start!

The instability problem remains no matter how long the unit is run for.

I do have 1 carbon resistor that measures 15K but the colors say 150R (the colors do not seem to be burnt) but from 150R to 15K seems a rather long way out to me.
This is a cathode resistor to 1 half of the ECC88, on the circuit I have it is shown as 15K (R28) it also shows the cathode going to all set frequency pots, in reality it does not, the set frequency pots are grounded. It does however go to C14 on the amplifier board as shown on the circuit diagram, but not to R66, R27 junction, they don't exist.

Finding the correct manual is pretty much order of the day now.

Dave

[davidgem1406]

A correction to my above posting (14).

it also shows the cathode going to all set frequency pots, in reality it does not, the set frequency pots are grounded.

I said the set frequency pots went to ground instead of the cathode, that is not correct, they do in fact go to the cathode. The connecting wire was threaded between a ground tag and a capacitor making it look to be connected to the tag.

Dave

[davidgem1406]

I have now got this working and stability is much improved, 15K carbon resistor replaced (cathode resistor to 1 1/2 of the ECC88). On the frequency counter the third decimal place digit moves up and down a bit, occasionally the second place decimal digit will move but that is dependent on the generator setting.

I did have the problem where the frequency for all bands would not reach the highest end and was some 20KHz to 30KHz short. I noticed previously that the preset capacitors for each top end had been padded out with mica capacitors although I had not taken any notice of this because of the circuit differences.

However I disconnected them and viola it was now only short at the top end to about 10KHz except the top band and that was a long way of by some 25KHz.

Mounted on the oscillator panel is a pot and according to the service info this is used to set the minimum distortion at 5KHz and with 25V of signal.
This pot is shown as being in the cathode of the ECF82 Pentode on the circuit diagram. In my unit it is not and is in the cathode of 1 half of the ECC88.
I adjusted this to see what effect it had and it increased the frequency to the 150KHz for the top end of the highest scale and no distortion that I could see on a scope. (The range of this generator is from 5Hz to 125KHz although the scale is marked to 150KHz) that now put the 125KHz to the right place on the scale.

At the higher frequencies I was getting a lot of muck on the waveform and assumed this was just pick up due to the oscillator panel just being strung in using longer wires so that I could get at the component side, but I felt that there was more to it than just that.

I decided it was a problem with the earthing creating an earth loop. The earth connections were rehashed and taken to the oscillator panel earthing points on the tag boards and one wire taken to ground. That cured the muck problem immediate.

All ranges are now close to their top and bottom ends. I will finalize all those and check other setting when I have resembled the oscillator and can refit all the screening. I have to wait for some 15K 2W resistors to arrive as I only had 1/2W available and that is getting to hot for comfort.

Still not been able to find any information for another version of the circuit for the SG66.
I have rough drawn the circuit of the oscillator section and will be making a good drawing of it, once done I will post both versions here, somebody may get a clue as to what went on here!

Dave

[XTC]

So you are making progress.

One of the things you are paying for with a commercial sig gen is for the earthing arrangements to have been sorted out at the R&D stage. I'm a little surprised that you can improve them.

If the resistors in the frequency determining network have crept high, the frequency would be lowered.

Usually, with a lamp or thermistor stabilised Wien Bridge oscillator the distortion adjustment is set so that the sig gen will oscillate reliably on all ranges. There is a very noticeable bounce on the lower ranges. If you set the adjustment to eliminate the bounce, you increase the distortion. Obviously, you follow the instructions in the manual if available. Some have a switched option where you can opt for low distortion using a thermistor, or fast settling and higher distortion (1 or 2%) using back to back diodes.

You probably won't notice distortion in a sine wave up to a few percent by using a scope. You have to use a distortion meter.


Pete.

[davidgem1406]

Hi Pete,

Yes good progress.

It does have a thermistor in the distortion circuit and the service instructions say how to set it. The problem is that my unit is not the same as the diagram and maybe the setup is different.

I will be continuing in the next day or so as I now have the 15K resistors. As for the frequency determining resistors I have not checked them but they are all good quality 1% or better resistors but if I still have problems when it is all back together I will go over them, although they would have to be a long way out to give an error of 20KHz to 30KHz.

No distortion meter available and not likely to be so I have to do whatever I can without it. The meter bounce is quite apparent at the low end of things.

Dave

[davidgem1406]

Here are the two versions of the oscillator 1 from the manual and the other as it actually is in the unit.

Comments are welcomed.

Dave.

[Radio Wrangler]

The really subtle bit about wien bridge oscillators is the amplitude stabilisation method. In both these diagrams a thermistor network is used to shunt (at AC at least) the anode load of an amplifier stage. Without the right thermistor, or with a damaged one the circuit runs away to silly gain and makes clipped waveforms, sometimes of odd frequencies.

The usual thermistor of this era was the STC R53.

In the era before, Bill Hewlett used a small panel lamp bulb and patented the thermitor stabilisation technique.

Variable-C tuning is very good. Double gang capacitors are a lot more predictable and better matched than available pots.

I've just picked up the later generation Advance wien bridge SG68A It uses a huge colvern triple-gang pot. It uses transistors and takes THREE PP9 batteries. Don't know if it works yet, I'm still saving up for the batteries!

David