Advance LF Generator Type SG66

[XTC]

It isn't a LED, it's a LED and Cadmium Sulphide Light Dependent Resistor combination. I 've seen factory made ones but you can make them up by taping a LED and LDR together, or sealing them in heat shrink. You have to keep out ambient light. Normal LEDS and LDRs you can buy in Maplins work well.

The point of the LDR is that at any given light level, it obeys Ohm's law. From what I recall from playing with one, the combination has a non-linear voltage/resistance curve like an NTC thermistor and some time delay, like the thermal inertia of the thermistor. The LED is usually fed by a precision full wave rectifier made from an op-amp or two and diodes plus some filtering.

http://www.redcircuits.com/Page82.htm

is an example of a circuit using it. It became popular in low distortion oscillators. It has nearly no bounce.

A FET was used in some commercial low distortion oscillator designs. However, at a given gate voltage, the FET drain-source current is linear over only a narrow range of voltage. It also has instantaneous response which is not what's needed, but that could be accommodated by using a capacitor to give a slower response.

I can dig out and scan some of the JLH articles describing this and giving practical designs..

Stabilising Wien Bridge oscillators is also discussed in detail in Linear Technology Application Note 43.

Starting page 29

http://cds.linear.com/docs/Application%20Note/an43f.pdf

Pete.

[Refugee]

XTC has the same idea as me.
The second link is very detailed and may confuse some readers as the OP amp is a high quality device and requires a lot of features to be set up to make it do what it is sold for.
We only need the basics though.
A low power OP amp with 1 ma or less consumption would work fine from the HT with a 150K resistor and Zena diode. This would make it a 3-wire device.
The simple one is the LED/LDR. This does not need any external power and may work as a 2-pin device with just 3 internal components.

[Refugee]

An analogue opto coupler might also work if anyone has had a dabble?

[XTC]

The LT application note is a bit OTT, and is bound to be something of a marketing tool for their products, but it covers quite a lot of relevant material.

I like the sense of humour the author has

Note 5: What else should be expected when trying to replace a single light
bulb with a bunch of electronic components? I can hear Figure 39’s #327 lamp laughing.

There's also an appendix (page 43) talking about Hewlett's original design.

TBH I think FETs and LED/CdS cells are something of a distraction and were it my problem I would have been inclined to mod the circuit and use a lamp or give the indirectly heated thermistor a go.

I doubt an optocoupler would do it. I can see problems with a LED/CdS cell without the extra circuitry but it would be cheap enough to try.

Pete.

[davidgem1406]

At the moment I am going with the heated thermistor, I just have to wait for them to arrive.
Once I have them I can see what happens. The heater will not be used just the thermistor as though it were an R53.

I have considered using the lamp option but I don't have the know how to be able to mod the circuit to accommodate that. I have looked at the HP circuits using lamp stabilization and they are a mile away from the Advance circuit. The lamp is in the cathode of one of the valves to ground, the Advance doesn't like that, I tried it.

Dave

[davidgem1406]

Im still trying to get this to work correctly. So this is the latest position.

I fitted the heated thermistor (without using the heater) and that unfortunately did not work at all well. The amplitude was quite unstable and the oscillator was reluctant to run on all wavebands.

I played with it for some time but in the end could not obtain any improvement. I also rewired again, back to the circuit as it was when I obtained the generator as this seemed to be the more stable circuit of the 2 but that also had very poor amplitude stability.

It seems the thermistor could not respond fast enough.

I will now try to find a way to mod the circuit so that I can use a lamp.
Failing that I am looking at replacing the existing oscillator circuit with something different, maybe from an HP circuit adapted to suit the frequency range required and to use B9A valves instead of octal.
I have collected several schematics, including 1 that uses a single Pentode for the wien bridge.

Solid state circuitry is not an option here as they all seem to use pots as the frequency selection and that is definitely not possible.

Dave

[Radio Wrangler]

Originally Posted by davidgem1406

Im still trying to get this to work correctly. So this is the latest position.
Solid state circuitry is not an option here as they all seem to use pots as the frequency selection and that is definitely not possible.

HP model 651A Wien bridge oscillator... all solid state, uses a tuning capacitor with fixed resistors. From the guy who invented thermistor stabilisation. I actually have a spare gang capacitor for an HP651A kicking about the place, but I can't remember the stabilisation method used. it may have been an FET as a variable resistance. The oscillator covered 10Hz to 10MHz and thousands were made. So there is a well proven circuit for a solid-state Wien bridge sitting in the manuals archive at either Agilent or BAMA.

Surest way to find something is to say it's impossible

Cheers
David

[WME_bill]

I am sorry to put a bit of a damper upon your struggles, and the interesting discussion, but you would make more progress is you could use a thermistor with a resistance and power rating comparable to the A15 (whether or not with a heater).
You appear to be comparing with a R53, rated 5kohm 3mw, frequently used in transistor Wien bridge oscillators. The other postings seem to be doing the same, which has perhaps misled you.
For you have a valve oscillator here at a greater power. The A15 specified by Advance is 100k, 60mw, and was used in most audio oscillators of that period. And not comparable at all to the R53. (Resistance is measured cold, at room temperature. Dissipation is usually max at 300degC.)
Scroggie, Radio Laboratory Handbook, most editions, has a discussion upon use of thermistors for amplitude control.
I see your B54 is rated 50k, 40mw, so that may work if you could reduce the power dissipation, and change the V2B anode load R23 so that the ratio R23:R26+Th1 makes sense for controlling the voltage fed on to the next stage V1B. (I assume you have the usual circuit references). An alternative is to increase in the series resistor R26, though this will reduce the effectiveness of the thermistor control of gain = amplitude

A Mullard alternative thermistor (1981) was VA3212 or similar. I see that RadioSpares is currently offering a resin encapsulated version made by EPCOS type S861 at about £1.50. They also list glass versions but they are only 32mw or 18mw, comparable to a R53, but too fragile for your circuit.
The resin version would do perfectly well within normal range of temperatures, though may be a bit slower to respond, due to larger thermal mass. However, should work quite well enough to get your unit working. Further improvements can follow later.
I have seen circuits in Wireless World using a FET or a light dependent resistor to control amplitude. Do you want me to search out the references.

The lamp idea was invented by Hewlett for the original HP200 audio oscillators of 1939. Made their fortune! Versions use 6v1w, or 250v10w bulbs, all at very low impedance between the cathodes of the amplifier valves, as tungsten bulbs have a positive temp. coefficient. It would need a major rebuild of your SG66 to find a suitable spot.

I believe EPCOS took over the former Philips Mullard business - can anyone confirm?
I attach extracts from a STC catalogue of 1967 giving specifications of the type A, B and type R thermistors. I have noted the AEI-BTH equivalent numbers which you sometimes see: STC A15 = TH-B15, and R53= TH-B18.
I also have details if anyone wants them of the other STC Thermistors and the Vacuum thermocouples which may be useful for meters in Marconi signal generators like the TF144 series, type TC-S.
Do keep us informed, as this is most interesting. WME-bill.

[davidgem1406]

Hi David,

So far I have not been able to locate the manual for the 651A, but the 652A seems to be similar. This uses a modified Wien bridge that uses a 3 gang variable cap for tuning.
However the SG66 has an output of 30V P t P and I'm not sure that a solid state circuit would be capable of driving the output amplifiers and the square wave circuitry to maintain that output. Also the additional power supplies required could be a problem as room on the chassis is very limited, it is quite well packed already.

Hello WME_Bill,
Thanks for your input and the information it contains. As you say the original thermistor was 100K at ambient temperature, until I destroyed it!!
Unfortunately these devices are no longer available thus my attempts to use other types that I was able to find, such as the B24 and the glass bead thermistor (cant remember the type number at the moment).
I did try increasing the value of the series resistor but if i remember correctly that did not help. Hence my comments about rebuilding to use a circuit that has a lamp as the stabilizer.

I could have a look at the thermistor from RS as I have an account there and maybe get one and try that. You said:

The resin version would do perfectly well within normal range of temperatures,

What would you consider the normal range of temperatures to be? as the unit when cabinet-ed up gets pretty warm with 8 valves running.

I assume the reference to using a FET is in regard to solid state circuitry. All the references I have seen to that say it leads to an increase in distortion.

I have the specifications of the type A, B and type R thermistors so have been using them as a cross reference one to the other but can find none of the usable thermistors available.

At the moment i am looking at maybe buying another SG66 then using the best parts from both to build just one good unit, not something I would normally think to do but sometimes when needs must!!
I have noted that all the J series units use Thermistor stabilizing most using 2 thermistors. There seems to be plenty of these units about but I cant find one that is a scrapper to take the thermistors from.

I will keep the thread updated with any progress.

Dave.

[davidgem1406]

Hello WME_Bill,

In your posting you recommended a thermistor EPCOS type S861 from RS.
This thermistor is only 3K @ 25 Deg. C
The Epcos B57861S is 100K 60mW @ 25 Deg. C, is this not the one that would suit best?

Thanks Dave

[WME_bill]

Thermistors. Yes, I must have misread the number. As you want one that is comparable to the STC A15 at 100k. Bill

[davidgem1406]

Thanks Bill

Dave

[davidgem1406]

I obtained a disk NTC thermistor 100K @ 25 C and fitted it.
It was a pretty poor response being very difficult and unreliable to oscillate on all bands, and virtually no stabilization of the amplitude.

At the moment I have taken an A15 thermistor from an Advance J2e to use.
It now oscillates reliably but again the amplitude stability is not good.
I now even refitted the oscillator panel back into place, shortened the wiring to remove the excess needed to be able to get at the underside.

That has now introduced parasitic oscillations siting on top of the sine wave. Why that has suddenly happened I have no idea as yet.
Also I noticed that the HT rail is, in my estimation, pretty dirty with 200mV of what looks like hum, but maybe it is correct as the HT is derived from a voltage doubler circuit so is only half wave rectified. That is without the oscillator running.

With the oscillator running there is far more rubbish on the HT rail especially the oscillator side of the 200R resistor (Rxx) particularly at the LF end of the frequency band where there is a significant amount of LF sine wave on the HT rail. I cant imagine that is correct .

I am now investigating further.

Dave

[davidgem1406]

I now have this unit working and oscillating reliably by using a Thermistor from an Advance J2E, I will worry about the J2E latter. Or I make a choice and sacrifice the J2E or an earlier J2, they each have the same type of Thermistor.

So far I have not bothered with setting the frequency ranges but they don’t look too far out.
The amplitude stability looks reasonably good on bands 3 to 5; with band 1 it is difficult to tell if there is some amplitude instability or not.
Band 2 looks as though it could have some amplitude instability but this is difficult to determine if it is actually an amplitude problem or if it is down to the problem I have with bands 1 and 2 as detailed below.

Also I checked the HT ripple at all points on the HT rail, I think it is quite reasonable for the type of HT supply being used (Voltage doubler so only half wave rectified).

Ripple check Without the oscillator running.

All voltages given in PtP.

1/. Ripple on the main HT rail (junction of R36 & R29) = 120mV

2/. HT rail @ point 33 (Schematic ref.) = 20mV
This rail shifts up and down by about 4mV and occasionally jumps by about 10mV.

3/. The oscillator HT rail point 17 (Schematic ref.) = <5mV has the same upward & downward shift as in 2/.

4/. The oscillator HT smoothing was increased to 150uF to obtain the figure as in 3/.


With the oscillator running

It was found that the low frequencies were beating with the mains supply frequency.
On band 1 between 10Hz and 49Hz, whilst 50Hz was steady. The worst point being 45Hz where the amplitude is modulated quite badly.
On band 2 frequencies below 50Hz and from 51Hz right up to 500Hz, the worst point being at 55Hz again the amplitude is modulated.

Adjusting the hum dinger pot RV8 on the heater supply does little to nothing to improve matters.

The original circuit contained another adjustable resistor RV15 and associated components R27 & R66, this was also used to minimize this beat on the output.
However the oscillator circuit as used in my unit does not have these components shown. In fact there is no position allocation for these components in my unit. Also the manual talks about the oscillator being on a Printed Circuit board, this is not the case for my unit.
The manual also mentions the adjustment of RV8 protruding from the screening at the rear, there is no allocation for this in the screening metalwork.

Other than the oscillator itself all other aspects appear to be identical to the circuit as shown in the manual, other than C27 being a trimmer, why and what for is another question.

Further progress to follow:

Dave

[davidgem1406]

Since my last posting I put the oscillator screening in place, this improves things quite considerably although I still have a problem with the sine wave beating quite heavily with the mains frequency at about 46Hz. The humdinger pot (RV8) does virtually nothing to help this problem.

Having the screening in place makes it difficult to work on the oscillator as it is now screening off, screening on, all the time.
The 150uf capacitor added to the oscillator HT supply I removed as this does nothing to improve the sine wave; the hum on the HT rail is within acceptable limits.

I have now also set RV6 to give 1.8V RMS across the 1K5 resistor in series with the Thermistor as per the manual.

The frequencies have been checked and set as best that is possible. I don’t think they are good. I spent a long time on band 5 trying to even out the results but what I did to the high end threw the low end out, correcting the low end then threw the high end out, it just went round and round.
This is the original problem that started this whole thread into being, if I remember correctly.

Frequency check

_____________________________Low end______High end
Band 1 ...... 5 to 50 Hz ............. 4 Hz ............. 50 Hz
Band 2 ...... 50 to 500 Hz .......... 47 Hz ........... 502 Hz
Band 3 ...... 500 Hz to 5 KHz ...... 490 Hz ......... 5.03 KHz
Band 4 ...... 5 to 50 KHz ............ 5.01 KHz ....... 50.24 KHz
Band 5 ...... 50 to 125 KHz ......... 54.32 KHz ..... 125.21 KHz

Band 1 continually changes between 4 and 5 Hz until the scale reads 5.2 Hz where it becomes a steady 5 Hz. So the 5 Hz is at the wrong point on the scale by 0.2 Hz. The 0.2 Hz equates to about 5mm of shift on the scale, quite a bit.

The rest of the setup is still to be done but that is just for the output metering and the square wave.

I now wonder if the best option is to rewire again, for the last and final time, to the oscillator circuit as in the manual and make the changes needed to accommodate the components that do not exist in the current circuit.

So far I still have no idea if the current oscillator circuit is a modification done by an individual for some reason or a factory built unit that is either earlier or latter than the manual that is available. No other information has come to light.

Maybe the original oscillator PC board had suffered some irreparable damage so somebody rebuilt using the layout that exists now i.e. a metal plate. However that does not give a reason to modify the circuit.

This saga has to come to an end one way or another as it just takes up to much time as I have other things I would like to be getting on with, an Advance SG63F and some radios for refurbishing for starters.

Dave

[davidgem1406]

The saga continues.

Recently I managed to pick up an Advance H-1 in poor condition with the sole intent to rob it of it’s Thermistor an STC A1522/100, just the job for the SG66. As a bonus I had been looking for a 6SN7 to keep as a spare for my Advance E2. The price they are asking for this valve is just pure robbery and greed. The H-1 had 2 of them, 1 very good & the other a bit down but usable, excellent.

I removed the Thermistor from the H-1 and fitted it into the SG66, the J2E can now have its Thermistor back.

The improvement to the SG66 was immediate and very apparent, I can only guess at the reason for this, the amplitude was stable, save for a small variation with some frequency changes. RV6 was set to an optimal point where oscillation on all bands was immediate with a good sine wave.

I was now able to set the top and bottom frequencies to within the +/- 1%, for the lower frequencies +/- 1 Hz, as specified on all but bands 1 & 2 at the low end of the ranges. To cure this I had to fit shunt resistors to the existing resistors R10 & R11. R10 shunted by 4M7 and R11 by 20M. All ranges were then set as follows:

Band............Range......................Low.... ..............High
1...............5Hz to 50Hz..............4Hz to 5Hz.........50Hz
2...............50Hz to 500Hz...........50Hz.................501Hz
3...............500Hz to 5KHz...........500Hz................5.016KHz
4...............5KHz to 50KHz...........5.014KHz............50.006KHz
5...............50KHz to 125KHz........50.029KHz..........125.126KHz

Band 1 slowly changes between 4Hz and 5Hz (Specification allows 4 to 6)

The sine wave is now all fully functional and looking good.

The adjustments for Output Impedance, Sine Wave Voltage, square wave Mark Space and Voltage have now been checked and adjusted. Some adjustments came right to the end of their available travel and only just made it.

The square wave is now the one with problems, this had not been checked before other than to know it was working.

Starting from the low frequency at 5Hz and amplitude of 25V P.t.P. the top and bottom are sloping by as much as 12V. (I have included a pictures but it is not the best as the camera could not sync to the scope display. Image 1 @ 10Hz and Image 2 @ 50Hz.)
The slope gradually reduces as the frequency is increased. At 500Hz it has all but disappeared although neither the top nor the bottom are absolutely flat.

From 500Hz to 25KHz it is reasonable from there on it degrades with the leading edges starting to round off. At 50KHz they are quite well rounded with a rise and fall time of about 2uS (Spec states 0.75uS max)
From there on it continues to degrade to where at 125KHz the top is completely rounded of with a rise time of some 3.5uS the fall time is still about 2uS.
Also the amplitude between 50KHz and 125KHz increases by 5V P.t.P. to 30V P.t.P.
Reducing the amplitude results in the waveform becoming more like a rounded saw tooth. (Picture added Image 3)

The square wave problem is caused within the Schmitt Trigger, Cathode follower or the Clipper Circuit, just 2 diodes back to back. The sine wave input has no effect on this fault condition other than the frequency range that produces the problem.
The square wave schematic is included. Voltage data for the valves is not given in the manual. The 2 valves have been changed with no improvement.

Any help or suggestions would be appreciated.

Thanks Dave

[Alistair D]

Dave,
I need to clarify a couple of points. Is the scope AC or DC coupled? What kind of oscilloscope probe are you using?

One of the possible sources of your problem could be the 100+60 uF electrolytic capacitor.

Al

[davidgem1406]

Hi Al,

No probe was in use it was a direct screen cable connection and AC coupled.
Using a probe x 10 gives the same result.

The waveform is the same either side of the 100+60 capacitor.

Dave

[ronbryan]

As Alistair has hinted, the scope should be DC coupled to correctly display a low frequency waveform, as the tilt in the waveform is caused by the low frequency roll-off of the scope AC input circuit.

To show higher frequency waveforms correctly you should use a proper scope probe, with the probe compensation capacitor adjusted on the scope 1kHz calibrator to give a good square wave with no tilt or overshoot.

Ron

[davidgem1406]

Hi Ron,

Thanks for the info i didn't know that at L.F. the scope needed to be DC coupled. Doing that shows the L.F. square wave is fine the sloping top an bottom don't exist.

At 50KHz up to 125KHz the same problem still exists but is not showing to be as bad.

I have a Farnell LFP1 so i did a comparison with the square waves just in case my scope or probe were producing the poor result. at %0KHz the Farnell shows a square waveform with just a small overshoot on the top edge. At 125KHz it degrades but is still a square wave and considerably better than the SG66. So from 50KHz up the SG66 waveform does have bad degradation, at 125KHz it is not now looking like a saw toot but has rounded edges and the rise and fall time is about 2us.

Dave.