Thanks, I remember reading he did that to prove TID/PID. It remains a truly remarkable example of independant thinking and just one of the reasons he is an audio icon but he is inaccessible to me via email or pm. If you have access to him, please forward my desire to ask him about this mod. Do you know the specs of the modified unit? If his mods can't get me a noise floor of -120dB with a 16V input from 5Hz to 50KHz, I may opt for something else.
You need to be more specific about your requirements. -120 dB from 16V would be 1.6 uV. A number of analyzers have a noise floor that low. What bandwidth? 30 KHz or 80 KHz? If you add distortion + noise there is only one commercial instrument: the APx555.
I believe a stock Boonton can just meet the noise requirement. That would be around 5 nV/rtHz in a 20 KHz band or the equivalent of a 2K resistor. I think even the APx555 is not much better than that.
Where does this requirement come from?
I believe a stock Boonton can just meet the noise requirement. That would be around 5 nV/rtHz in a 20 KHz band or the equivalent of a 2K resistor. I think even the APx555 is not much better than that.
Where does this requirement come from?
Yes, 16V at 1ppm would seem to be 16uV so that relaxes the requirement slightly and even though I stated the upper bandwidth requirement at 50KHz, I could live with 30KHz. If I'm not mistaken the HP8903A fails to meet this spec at this noise floor but anyone with ideas of what I should be looking at, please sound off. Any Quantech gear meet my spec requirement? Any modded gear?
Seems I missed a decimal point. 16 UV in 30 KHz is easier. It's still very low and usually the test setup limits the measurement with hum and noise pickup. I think you could get there with an ST1700. JC upgraded his with a few AD797's. You need to find adapters from dip to TO8, they are available. The bigger issue with the ST boxes is the age. If they have seen a lot of use then they can be troublesome. I have 3 but would not depend on them professionally.
Quantech never made audio analyzers. The 8903 is pretty good. The specs are real and most will do beter. The noise is limited by three factors. 1) input amp noise. 2) attenuator impedance and 3) protection circuit. The protection circuit is the biggest problem. If you defeat it it's very easy to fry a number of delicate parts.
My understanding from JC was that it was opamp upgrades and careful alignment. There are a few tweaks that can improve its performance. There are better optocouplers than were used when the ST stuff was made. The Lightspeed attenuator stuff uses them.
Here are some old notes from me http://www.diyaudio.com/forums/equi...ound-technology-st-17xx-gear.html#post2228938
The bad news is that the Silonex stuff that was the core of the opto attenuators has been killed off and the company, Silonex, sold to another outfit that doesn't make the stuff anymore. Advanced Photonix, Inc. | Optocouplers There is a lot of hype on the site but no products. I should get some just for maintenance if we can figure out who has usable parts.
Here are some old notes from me http://www.diyaudio.com/forums/equi...ound-technology-st-17xx-gear.html#post2228938
The bad news is that the Silonex stuff that was the core of the opto attenuators has been killed off and the company, Silonex, sold to another outfit that doesn't make the stuff anymore. Advanced Photonix, Inc. | Optocouplers There is a lot of hype on the site but no products. I should get some just for maintenance if we can figure out who has usable parts.
@dellama -- no matter how many times you ask, you probably won't get an answer -- not being snarky here, it's just that nobody knows. Depending on what you're doing, I can suggest using an audio range spectrum analyzer combined with an active Twin-T filter and attenuator front end. Resolving power of harmonic products is demonstrated to be below -130dBu... There's a forum page here for this topic, called Active Twin-T Filter IIRC.
Thanks, but that still doesn't tell me how low it can measure after the mods
When JC was asked about thie mods he said they got the noise down another 10dB. There was no mention of a reduction in distortion measurement. Sound Technology recommends using a notch filter to increase the resolution from 0.002% to 0.0002%.
The oscillator in the 1700 series is the weak spot in those analyzers. The switches are forever causing problems. The analyzer is great.
if your measuring noise its one problem and not too hard to solve. There are a number of microvolt resolution audio frequency voltmeters that will do the job. I just confirmed that a Boonton 1120 has a noise floor that is -120 dB from 16V. Measuring -120 dB THD+N is another much harder task, as I pointed out before.
You could use one of these B K Bruel Kjaer Type 2609 20 20000 Hz Industrial Measuring Amplifier | eBay to measure the noise. it works quite well and is a nice little box. i have 2 I use with the B&K mike preamps and microphones for acoustic measurements.
You could use one of these B K Bruel Kjaer Type 2609 20 20000 Hz Industrial Measuring Amplifier | eBay to measure the noise. it works quite well and is a nice little box. i have 2 I use with the B&K mike preamps and microphones for acoustic measurements.
That will take more than a twin T filter. It took Bruce Hofer 10 years and lots of resources to come up with all the tricks he used. You may be able to get there with Davada's efforts but that is unknown at this point.
Just get the APx555. The chart on the first page http://www.ap.com/download/file/758 will give you a relative comparison. Even a hand tweaked Shibasoku is not that good.
Other than bragging rights there is not a lot of need to go beyond -100 dB, -110 if you are dealing with SOTA data converters or characterizing opamps.
Hi Demian,
I really need to do something about my day job! I think I should have some time to spend on this again this Thanksgiving weekend. Sorry that the multiplier mods have not worked out for you. I was wondering if you would see any improvement when you already had residuals at 0.0005% without it. I did not expect it to go in reverse. I was seeing 0.0015% before the modifications so I was pleased to think I had it licked. I need to rethink the multipliers. I am trying to stay within the constraints of making simple modifications first that others can try easily. I went further down the road of attenuating the auto-tune and auto-balance feedback loops to where they maintained the null and contributed as little as possible to the residuals. I got THD+N in the mid 0.0003% range. Believing that the multipliers are still the primary source of distortion and given that the multiplier is obsolete, I am considering making a daughter board that pugs in their place with a FET-based design that is so popular on other analyzers. Any suggestions from the community is appreciated. This should improve the source side as well. I will also look again at the RC4200 data sheet and see if I can understand why yours is providing the opposite effect as the three boards I have tried. There must be another factor here that I am missing. This is truly like peeling an onion!
Glad to hear that the nulling is working better. That one always annoyed me, both from a residual drift perspective and from a feed through at higher frequencies. At least it is progress. Thanks for being patient. I will let you know what I find.
Mike