How Distortion Free are the Distortion Measurers?

[klewis]

Originally Posted by richiem

I think carbon film resistors get a bad rap concerning noise and other issues -- in most circuits they work perfectly well, so long as you know that they have larger temperature and other stability drifts, and use better resistors in critical paths. If gain or amplitude stability is important, use metal film or wirewound (but watch that L/R factor); if frequency stability is important, etc.

Hi Dick,
As I mentioned to Demian, I built a spare Amp 1 board, seem to work fine in testing, but one odd thing happened. When I hooked it up to your twin t at 1vrms out, the null dropped to about -80db. I've never been able to get much more that -60db on a really good day. Though I haven't measured it, I believe that the Amp 1 has a big DC offset that is taken out by a 10uF cap in the next stage, but don't know why I would get such a big null when testing this amp through your twin t.
Ken

[1audio]

Check that the twin tee isn't loading the amp in a way that causes a lot of the drop. The Quantech has a notch filter as part of the AGC circuit and "compresses" the gain to meet a target level for the reference tone. The output Z may be quite high and the notch filter actually causes a drop at the notch frequency.

I'll scan the compressor circuit tonight. I have several I'm (not) trying to maintain. More insight always helps.

[richiem]

The input Z of the Twin-T varies a lot depending on frequency. Might be a good idea to monitor the input level to the T-T while you have it nulled, as Demian pointed out -- if the level drops when you switch the T-T in, then you have the explanation.

I've resisted putting a unity-gain buffer in front of my T-T in order to minimize distortion, but that could be a good idea here. I would use an OPA1641 were I to do it.

[klewis]

Originally Posted by 1audio

Check that the twin tee isn't loading the amp in a way that causes a lot of the drop. The Quantech has a notch filter as part of the AGC circuit and "compresses" the gain to meet a target level for the reference tone. The output Z may be quite high and the notch filter actually causes a drop at the notch frequency.

I'll scan the compressor circuit tonight. I have several I'm (not) trying to maintain. More insight always helps.

Demian and Dick, Thanks for the helpful suggestions. I'll follow Dick's suggestion for voltage droop. I am currently testing the Amp 1 board seperated from the Quantech, i.e. I bought a socket and have rigged up a power supply and input / outputs to test it stand alone inside of a metal box. A bit of overkill, but, I love projects...
Ken

[klewis]

Test of V2 of Amp 1. both tests are with 1vrms output, no voltage droop when connected to Dick's twin t, but, upon the second test the null dropped to -80 or so. Wierd! On second test I connected the amp 1 output to the Hp 8903 to get the Vrms reading, that's when the null dropped. But, when I disconnected the 8903 the null didn't come back up.

Any thoughts?

Thanks

Ken

[jackinnj]

I got this from a company in Michigan that was going out of business -- also a bunch of Hewlett Packard power supplies:





This block diagram might be helpful:



When I get around to it I'll scan the manuals.

WRT Carbon Comps - they have to be derated at V>200.

[klewis]

Originally Posted by jackinnj

I got this from a company in Michigan that was going out of business -- also a bunch of Hewlett Packard power supplies:


When I get around to it I'll scan the manuals.

WRT Carbon Comps - they have to be derated at V>200.

Hi Jack,

Much newer than the one I have. I have the 2173C which is the module that accepts the transistor, biases it and creates the noise signal. It's output is then connected to a 2181 which has amps, bandpass filters, and meters to display the noise output. I built the 2181 from the schematics... but only the 10hz and 1khz sections as they seem most relavant.

Ken

[1audio]

I also have a 5173. Did you get any of the socket adapters? Mine came with one and I built another. My manual has no documentation on the adapters. I would really like to figure out one for dual transistors but I see no way to implement it in the box.

You can use one of the classic LED spectrum analyzers (eBay. . ) with the 2173. I did originally eons ago. Calibrating it is confusing however. The noise generator is key to getting everything in line.

[klewis]

Originally Posted by 1audio

I also have a 5173. Did you get any of the socket adapters? Mine came with one and I built another. My manual has no documentation on the adapters. I would really like to figure out one for dual transistors but I see no way to implement it in the box.

You can use one of the classic LED spectrum analyzers (eBay. . ) with the 2173. I did originally eons ago. Calibrating it is confusing however. The noise generator is key to getting everything in line.

Hi Demian, when you say "classic LED spectrum analyzers" any particular brand? or model number?

Any luck scanning the compressor schematic from your old manual?

I was in a bit of a hurry this morning, at this point, I'm pretty confidant that my duplicate amp 1 board works well. Still have to check it's self noise, probably won't get to that until next week, going out of town for a few days.

Just a bit stumped by this 40db drop in the fundamental...

Ken

[klewis]

If you guys are up for it, I have a request... first a little background on the quantech 2181. The schematics show all the amp circuits, and for the the 10khz section the bandpass filter is made with resistors, caps and transistors. But, for all the other frequencies, the bandpass filter also involves inductors that are not specified. So, I have no idea of the bandpass width for the other frequencies. I you had a spare moment and could measure the bandpass at 100hz and 1khz of your quantech machines, I could be sure that my design matches or is close to the original. For my current 1kHz bandpass I'm using 950hz to 1050 hz as the minus 3db points which give me a flat top from about 970 to 1030 hz.

Ken