Solartron OS101 signal generator strange "calibration". Any ideas? - UK Vintage Radio Repair and Restoration Discussion Forum

Jez1234 · 13th Sep 2023, 11:45 pm

I'm restoring this old sine wave generator and there are some strange aspects to it's calibration etc that have me really scratching my head!

I wonder if the following will sound as odd to others as it does to me. The instruction manual (I haven't got and can't find on line a service manual but the one I have has parts list and schematic. Just no calibration procedure so I'm having to make it up as I go along) states the max output as "100mW into 600R". Now this is of course 7.75V and with the unit set to max output it does indeed indicate precisely 7.75V on its own meter. There is in fact a calibration point at precisely this output with a (very faded) red line on the meter as the cal point.

The problem is that two accurate true RMS DMM's and an oscilloscope all say that with 7.75V indicated on the built in meter the output is in fact about 11.7V RMS! The built in meters' max reading is 10V.

It gets stranger though! I'm reasonably happy that I've ruled out any fault with the meter (see other thread on Weston Crystal Rectifier panel meter).

The original series SOT calibration resistor provided with the meter is marked 19.8K. It actually reads 20.1K so it has drifted a tiny bit high but still reasonably accurate. I've added my own shunt resistor across the original one to obtain a true accurate reading on the meter and have ended up with a total of approx 13K (12.88K). A measured 7.75V RMS AC now indicates correctly on the meter. Calibration now proves correct at all indicated points on the built in meter as checked on accurate external meter.

Now the thing is though that I've kinda ruled out any fault with the unit (it worked from being built up on variac for a few hours and various minor faults have been fixed now) hence it appears to have been designed to show 7.75V when the actual output is around 11.7V!!!

I don't get it.... Further "evidence" is that the internal output level calibration control only covers around 9.5 - 12.5V RMS range with the front panel level control set to max! The internal level control functions by varying the current through the thermistor as one would expect. One could suspect that the thermistor is out of tolerance but the front panel meter has been shown to not be faulty and yet it (before my little mods) reads 7.75V when true output is 11.7V ie within the range of adjustment provided.... and the unit cannot be adjusted to give as low an output as 7.75V with front panel control at max.

There is a front panel switch to put 600R across the output and all works as predicted with this. To be clear, the Wien bridge oscillator is followed by a feedback amplifier with an output Z of a pretty low 10R. The meter reads the output at this point before the 600R attenuator. The attenuated output reads correct ie twice what it should be into open circuit and correct into 600R with a decent amount of attenuation selected ie 20dB

This is all bizarre to me as it seems it's meant to be like this! EH? The alternatives are weird such as identical output meter but with wrong ballast resistor has been fitted at some point AND the thermistor has drifted such that 7.75V RMS cannot be obtained (min about 9.5V RMS).. but it appears completely unmolested in these areas and the red dye stuff on original solder joints was all present and correct.

Anyone got any ideas on this one??

Radio Wrangler · 14th Sep 2023, 5:46 am

If it's got a 600 ohm attenuator, switched, at the output which can be switched all the way out, then it would only make sense if the feedback amp output Z was always padded out to 600 ohms. Otherwise the output Z presented to the load would jump between attenuator out and any other attenuator setting, so the output level accuracy would be affected bu load impedance AND attenuator setting, or the effective attenuator value would be differently wrong with different impedances.

Solartron put plenty of effort into their stuff, but they didn't always make things which worked in the mainstream way. Some of their stuff was before a mainstream approach had established itself.

David

Jez1234 · 14th Sep 2023, 11:41 pm

Yes this also struck me as a bit odd but it is indeed how things are. There is an output from before the attenuator, the level of which is simply controlled by the front panel output level control. This has an output Z of 10R from the buffer (the first two valves, an EF91 and an N78, form the Wien bridge whilst the second two, same as the first pair, form a buffer with 100% feedback) and also feeds the attenuator. With the attenuated output set to zero attenuation it simply feeds the main output at 10R output Z to the attenuated output terminals. However, I've just run tests on this and with the internal precision 600R resistor switched in the attenuated output is spot on at all settings of the switched attenuators, which allow up to 60dB attenuation in 1dB steps.

With further thought on the subject I have to conclude that the meter had lost loads of sensitivity somehow... a fading magnet maybe? It was out by a mile. Reducing the ballast resistor from 20.1K to 12.88K has resulted in excellent accuracy from the meter.

The schematic and parts list mentions a "10V AC 2-1/4" " meter and mentions the "resistor, provided" so I'm as confident as I can be that the meter is indeed a simple 0 - 10V AC RMS reading instrument and the shenanigans I've had were due to the sensitivity dropping by something like 40% over time.

As to the output level well that's less clear cut but I don't think it's important really. All components in the oscillator check out as OK and within tolerance (I replaced a few that didn't previously) and changing out the EF91 makes no difference, as expected. The only thing left that could possibly cause the output level to be too high would be the thermistor drifting high in value. If this were the case then I could of course restore the correct output level by increasing the value of the lower arm feedback resistor to compensate. I'm unconvinced though that there is really any problem here. I do wonder if when it says "Max output 100mW into 600R", hence 7.75V RMS, what it really means is that the variable output level control should be set to this level and no more in order for all specs to fully be met or some such?....and that yes you could set it for say 11V or whatever? Without a proper service manual I'll never know I guess. Anyhow as the meter reads to 10V and I like the concept of decimal I've preset a max output of 10V which can then be turned down to 7.75V as required using the front panel output control. It measures an impressive for the time 0.2% THD so this output seems fine.

13th Sep 2023, 11:45 pm	#1
Jez1234 Banned Join Date: Mar 2021 Location: Morpeth, Northumberland, UK. Posts: 936	Solartron OS101 signal generator strange "calibration". Any ideas? I'm restoring this old sine wave generator and there are some strange aspects to it's calibration etc that have me really scratching my head! I wonder if the following will sound as odd to others as it does to me. The instruction manual (I haven't got and can't find on line a service manual but the one I have has parts list and schematic. Just no calibration procedure so I'm having to make it up as I go along) states the max output as "100mW into 600R". Now this is of course 7.75V and with the unit set to max output it does indeed indicate precisely 7.75V on its own meter. There is in fact a calibration point at precisely this output with a (very faded) red line on the meter as the cal point. The problem is that two accurate true RMS DMM's and an oscilloscope all say that with 7.75V indicated on the built in meter the output is in fact about 11.7V RMS! The built in meters' max reading is 10V. It gets stranger though! I'm reasonably happy that I've ruled out any fault with the meter (see other thread on Weston Crystal Rectifier panel meter). The original series SOT calibration resistor provided with the meter is marked 19.8K. It actually reads 20.1K so it has drifted a tiny bit high but still reasonably accurate. I've added my own shunt resistor across the original one to obtain a true accurate reading on the meter and have ended up with a total of approx 13K (12.88K). A measured 7.75V RMS AC now indicates correctly on the meter. Calibration now proves correct at all indicated points on the built in meter as checked on accurate external meter. Now the thing is though that I've kinda ruled out any fault with the unit (it worked from being built up on variac for a few hours and various minor faults have been fixed now) hence it appears to have been designed to show 7.75V when the actual output is around 11.7V!!! I don't get it.... Further "evidence" is that the internal output level calibration control only covers around 9.5 - 12.5V RMS range with the front panel level control set to max! The internal level control functions by varying the current through the thermistor as one would expect. One could suspect that the thermistor is out of tolerance but the front panel meter has been shown to not be faulty and yet it (before my little mods) reads 7.75V when true output is 11.7V ie within the range of adjustment provided.... and the unit cannot be adjusted to give as low an output as 7.75V with front panel control at max. There is a front panel switch to put 600R across the output and all works as predicted with this. To be clear, the Wien bridge oscillator is followed by a feedback amplifier with an output Z of a pretty low 10R. The meter reads the output at this point before the 600R attenuator. The attenuated output reads correct ie twice what it should be into open circuit and correct into 600R with a decent amount of attenuation selected ie 20dB This is all bizarre to me as it seems it's meant to be like this! EH? The alternatives are weird such as identical output meter but with wrong ballast resistor has been fitted at some point AND the thermistor has drifted such that 7.75V RMS cannot be obtained (min about 9.5V RMS).. but it appears completely unmolested in these areas and the red dye stuff on original solder joints was all present and correct. Anyone got any ideas on this one??

14th Sep 2023, 5:46 am	#2
Radio Wrangler Moderator Join Date: Mar 2012 Location: Fife, Scotland, UK. Posts: 22,262	Re: Solartron OS101 signal generator strange "calibration". Any ideas? If it's got a 600 ohm attenuator, switched, at the output which can be switched all the way out, then it would only make sense if the feedback amp output Z was always padded out to 600 ohms. Otherwise the output Z presented to the load would jump between attenuator out and any other attenuator setting, so the output level accuracy would be affected bu load impedance AND attenuator setting, or the effective attenuator value would be differently wrong with different impedances. Solartron put plenty of effort into their stuff, but they didn't always make things which worked in the mainstream way. Some of their stuff was before a mainstream approach had established itself. David __________________ Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done

14th Sep 2023, 11:41 pm	#3
Jez1234 Banned Join Date: Mar 2021 Location: Morpeth, Northumberland, UK. Posts: 936	Re: Solartron OS101 signal generator strange "calibration". Any ideas? Yes this also struck me as a bit odd but it is indeed how things are. There is an output from before the attenuator, the level of which is simply controlled by the front panel output level control. This has an output Z of 10R from the buffer (the first two valves, an EF91 and an N78, form the Wien bridge whilst the second two, same as the first pair, form a buffer with 100% feedback) and also feeds the attenuator. With the attenuated output set to zero attenuation it simply feeds the main output at 10R output Z to the attenuated output terminals. However, I've just run tests on this and with the internal precision 600R resistor switched in the attenuated output is spot on at all settings of the switched attenuators, which allow up to 60dB attenuation in 1dB steps. With further thought on the subject I have to conclude that the meter had lost loads of sensitivity somehow... a fading magnet maybe? It was out by a mile. Reducing the ballast resistor from 20.1K to 12.88K has resulted in excellent accuracy from the meter. The schematic and parts list mentions a "10V AC 2-1/4" " meter and mentions the "resistor, provided" so I'm as confident as I can be that the meter is indeed a simple 0 - 10V AC RMS reading instrument and the shenanigans I've had were due to the sensitivity dropping by something like 40% over time. As to the output level well that's less clear cut but I don't think it's important really. All components in the oscillator check out as OK and within tolerance (I replaced a few that didn't previously) and changing out the EF91 makes no difference, as expected. The only thing left that could possibly cause the output level to be too high would be the thermistor drifting high in value. If this were the case then I could of course restore the correct output level by increasing the value of the lower arm feedback resistor to compensate. I'm unconvinced though that there is really any problem here. I do wonder if when it says "Max output 100mW into 600R", hence 7.75V RMS, what it really means is that the variable output level control should be set to this level and no more in order for all specs to fully be met or some such?....and that yes you could set it for say 11V or whatever? Without a proper service manual I'll never know I guess. Anyhow as the meter reads to 10V and I like the concept of decimal I've preset a max output of 10V which can then be turned down to 7.75V as required using the front panel output control. It measures an impressive for the time 0.2% THD so this output seems fine.