I agree with everything you say, but the EM electronics didnt get much mention first time around either: http://www.diyaudio.com/forums/anal...rch-preamplifier-part-ii-219.html#post2498560
Wrinkle
It is my opinion that people should read the textbooks written by Holm or the papers on contact problems written by J.H Whitley, if they want to really understand what happens with contacts, especially over time, vibration and temperature. Of course, a FRESH contact of just about any metal should be OK for awhile, but with time, things change.
Both Charles Hansen and I selected Shallco silver on silver switches because we found that they sounded the most neutral of any contact that we could find.
Personally, I have designed amps and preamps with: relays, both sealed and open, solid state switches, both integrated and discrete, connectors, sometimes gold plated, sometimes not. Yet, I have found silver on silver switches to be the most transparent.
I have found distortion in rotary potentiometers, some of very high cost, yet they had AUDIBLE distortion at certain settings. Where do you find this in any handbook? Perhaps, one should sometimes learn from the experience of others.
Both Charles Hansen and I selected Shallco silver on silver switches because we found that they sounded the most neutral of any contact that we could find.
Personally, I have designed amps and preamps with: relays, both sealed and open, solid state switches, both integrated and discrete, connectors, sometimes gold plated, sometimes not. Yet, I have found silver on silver switches to be the most transparent.
I have found distortion in rotary potentiometers, some of very high cost, yet they had AUDIBLE distortion at certain settings. Where do you find this in any handbook? Perhaps, one should sometimes learn from the experience of others.
John, thanks for sharing many of your insights. I believe in the Stradivarius effect. Brilliant things can come about when all big things and all details are implemented just right, even if done only by ear because measurements cannot be done or are inconclusive.
But, the Timothy Leary thing "question authority, think for yourself" has hit hard with me. I make some money in the audio industry through a licensing agreement, yet I am not a specialized audio electronics engineer. However, having done some complicated electronic designs in other fields, I am amazed by many of the claims made by people I should consider excellent, even driven engineers, based on their history in this field and the products they have brought to the market.
It still befuddles me. In other fields of engineering they would be launched out of the window, not just the claims, but the engineers as well.
But, the Timothy Leary thing "question authority, think for yourself" has hit hard with me. I make some money in the audio industry through a licensing agreement, yet I am not a specialized audio electronics engineer. However, having done some complicated electronic designs in other fields, I am amazed by many of the claims made by people I should consider excellent, even driven engineers, based on their history in this field and the products they have brought to the market.
It still befuddles me. In other fields of engineering they would be launched out of the window, not just the claims, but the engineers as well.
Last edited:
Attached is my estimate of how to build a pico voltmeter.
You adjust R1 until there is no voltage out from the amplifier. A vary narrow bandpass filter or other noise reduction method may be used to look at the output voltage. When the output is 0 volts the meter will read 1,000,000 times the input. Note that the reference voltage and the signal under test are both the same polarity.
You may need to parallel relays and amplifiers to get lower noise and faster settling time.
This circuit will show the rising input impedance as in the spec's and also the long settling time on low level signals.
You adjust R1 until there is no voltage out from the amplifier. A vary narrow bandpass filter or other noise reduction method may be used to look at the output voltage. When the output is 0 volts the meter will read 1,000,000 times the input. Note that the reference voltage and the signal under test are both the same polarity.
You may need to parallel relays and amplifiers to get lower noise and faster settling time.
This circuit will show the rising input impedance as in the spec's and also the long settling time on low level signals.
Attachments
I had a 5 digit DVM from the early 60's that used a mechanical chopper, tubes, and phone exchange switch/relays to "dial-in" the voltage that worked like that. Amazing ingenuity. The thermal problems at 1pV are huge, you also have several mechanical connections. 😀
Yes, all the tough considerations at a picovolt level are basically thermal. I can find reed relays that will do the job. (Although you do have to use them in pairs to really reduce their thermal issues, but that has been covered elsewhere.)
I think without tweaking you could make this work around 30pv! 1 nv/hz at bw 1/300 = 58pv 8 pairs of differential relays (1 amp per) ~ 21 pv
Last edited:
Yes, besides the point made by Scott, there might be a fundamental problem with this setup. If I calculated right, only perhaps one or two electrons per second would enter the device at 5pV over 10Mohm. When switching at 257 c/s, good luck catching one at the right moment. Another is that the resistor connecting the flying cap to earth should be very, very large = lots of Rn on top of a very, very weak signal. At the danger of looking like Monty Python's Spanish Inquisition, three, you would still be pretty much limited by the noise level of Vout. How to filter out from this soup of dancing electrons a signal consisting of only a couple of electrons per second? It is the same problem you started to begin with, compounded by, four, the problem that you are bleeding current into the input when R1 is on the high side.
vac
Last edited:
You really don't understand it! You should have no input current when it is in balance. The current bleeding out when the references is high is part of the process.
Sure I understand as I was quite precise in my formulation (when R1 is on the high side).
I would not want a pV measuring device to throw any current into my sample at any time. This is unavoidable in this implementation. Some samples like thermocouples are robust. Others may not be.
Another way of looking at it is that this setup will have a very low negative inverse input impedance during parts of the measurement.
Last edited:
I doubt without a source reversal differential measurement you could do better than ~ micro-volts measurement.
Lots of good stuff here, a bible as it were.
http://www.keithley.com/knowledgecenter/knowledgecenter_pdf/LowLevMsHandbk_1.pdf
Lots of good stuff here, a bible as it were.
http://www.keithley.com/knowledgecenter/knowledgecenter_pdf/LowLevMsHandbk_1.pdf
E-6 vs E-12. Which means that the distance between 1V and what you can achieve with differential measurement, is the same as between differential measurement and what EM achieves. On a log scale. Quite a mystery still how they do it.
I'm using this little circuit as an output buffer in my current phono equalizer project, but thought that it might be suitable as a poor man's Blowtorch substitute. For those of us who use all single-ended sources (or push-pull sources as two polarity choices of the same thing) and can live with gains of plus and minus one, a paraphase does the trick, and no loop feedback.
Figure 2 shows the original split-load, "concertina" inverter, and figure 1 is its JFET folded cascode version. Equal R's give equal magnitude outputs. For line output use the current sources can probably be resistors, and scaling for +10dBu max outputs and 1000 ohm loading R's should allow inexpensive FET pairs like 2N5459/2N5462 or J111/J174. With lots of hand matching, natch, but 100-lots from Mouser are do-able. As JC says, buy your lifetime needs now.
Not a real Blowtorch, of course, but maybe a reasonable DIY version. Comments always welcomed. Thanks, Chris
Figure 2 shows the original split-load, "concertina" inverter, and figure 1 is its JFET folded cascode version. Equal R's give equal magnitude outputs. For line output use the current sources can probably be resistors, and scaling for +10dBu max outputs and 1000 ohm loading R's should allow inexpensive FET pairs like 2N5459/2N5462 or J111/J174. With lots of hand matching, natch, but 100-lots from Mouser are do-able. As JC says, buy your lifetime needs now.
Not a real Blowtorch, of course, but maybe a reasonable DIY version. Comments always welcomed. Thanks, Chris
Sorry for quoting myself, but couldn't edit the above post anymore.
- to put this enigma in the right perspective (because a log scale does not do it right) let's look at it on a linear scale: if the Sun would be at 1 Volt, and the Earth would be at 1 uV, EM's machine could measure 34.223.100 light years away, some 2-3 times more than the age of the Universe allows.
That requires new technology.
- Status
- Not open for further replies.