When Pete Stockman was being considered for a faculty position at UCLA he chatted with the grad students (I was sometimes allowed to participate despite being an employee) and said that at University of AZ and Kitt Peak he advocated balanced (he called it differential) everywhere. I raised the objection of worse intrinsic/stochastic noise and he asserted that it was ALWAYS negligible compared to interference effects at the telescope.
When Dr. Shelton took over the development of an infrared speckle interferometer from me when I left the department out of frustration, he also insisted on the same philosophy more-or-less. Yet, I got good results with a single-ended approach in many cases, but with attention to details that others might brush aside.
The one environment I never had a chance to test the one instrument in was Mt. Wilson, home to a flock of TV transmitters. Now that place was fierce. Unfortunately a ceramic disc cap in the bowels of the instrument decided it was its time to start to fail, and it took weeks to diagnose. I was actually teased by one of the lackeys for my principal enemy in the department after this, and came within a hair of attempting to kill him. It takes a lot (so far there has never been enough) to move me to violent acts.
When Dick Aikens and his company contracted to build a detector system for Ed Rhodes to be used at Mt. Wilson for solar astronomy, I told Ed to stipulate that it would meet its performance specs in the Mt. Wilson environment. Of course it wasn't even close, and for the favor of showing them how to get within about a dB (not the units used but to give an idea) of the Arizona lab performance, I was promptly hated by the resident technician Maynard Clark, and UCLA was supposedly characterized as doing things cheap and dirty (I indicated where proper shielding should be fabricated using aluminum foil, then failed to make the recommendation in writing after leaving). What I did in essence was to make a waveguide-like trap as a common-mode choke, as one of the primary remedies.
In the modern land of ubiquitous email, this sort of political fecklessness would be avoided today. I was naive, to say the least.
When Dr. Shelton took over the development of an infrared speckle interferometer from me when I left the department out of frustration, he also insisted on the same philosophy more-or-less. Yet, I got good results with a single-ended approach in many cases, but with attention to details that others might brush aside.
The one environment I never had a chance to test the one instrument in was Mt. Wilson, home to a flock of TV transmitters. Now that place was fierce. Unfortunately a ceramic disc cap in the bowels of the instrument decided it was its time to start to fail, and it took weeks to diagnose. I was actually teased by one of the lackeys for my principal enemy in the department after this, and came within a hair of attempting to kill him. It takes a lot (so far there has never been enough) to move me to violent acts.
When Dick Aikens and his company contracted to build a detector system for Ed Rhodes to be used at Mt. Wilson for solar astronomy, I told Ed to stipulate that it would meet its performance specs in the Mt. Wilson environment. Of course it wasn't even close, and for the favor of showing them how to get within about a dB (not the units used but to give an idea) of the Arizona lab performance, I was promptly hated by the resident technician Maynard Clark, and UCLA was supposedly characterized as doing things cheap and dirty (I indicated where proper shielding should be fabricated using aluminum foil, then failed to make the recommendation in writing after leaving). What I did in essence was to make a waveguide-like trap as a common-mode choke, as one of the primary remedies.
In the modern land of ubiquitous email, this sort of political fecklessness would be avoided today. I was naive, to say the least.
An Orange Roughian perhaps. And violin acts --- I'd have better luck with a viola. Around here, all you have to do is brandish one. Critics who wish to control and ultimately confiscate all violas (except those played by Homeland Security) fail to take the deterrent effects into account, and only cite the murders and broken homes from domestic viola violence.
What Bill told me, when we were discussing his circuit, was that transformers do not work well at higher frequencies, where imbalance in windings becomes difficult to control. That was the reason a transformer manufacturer would be making an electronic equivalent to his own product. The core issue is common mode to differential mode conversion. Any impedance mismatch to ground on either side will will contribute to the conversion. The benefit of a center tap is a lower impedance source (hopefully) making the conversion smaller.
I wish the common mode was that good. I worked with Jensen on this exact issue for an isolation transformer for subwoofers. The common mode hum transfer was good but not always good enough, even with shields etc. When there was no real problem they worked great. With as little as 60V common mode the hum was no longer acceptable. I learned a lot from that exercise. The transformer is now in their catalog I believe.
That is the real issue, further issue is RFI coupling through the transformers - they WORSEN the situation.
The V to I conversion of any bipolar is essentially the same and the FET's are always much better, but I must remind you again that the collector current is related to Vbe (at 0 Vce) by the diode equation regardless of beta. You can make log amps with very low beta transistors limited by parasitics like big RE and rbb.
Last edited:
I was just back at my bench working on exactly that problem! For fun we have been building an assortment of microphone preamplifiers. The first one is a a single op amp gain stage and a dual opamp servo. Transformer input at 1/2 ratio. Local AM station came in at -50 (re 1 V). Ah the reference designs from some manufacturers!
The power supply is in a steel box and connected by a simple 4 conductor cable (+15, -15, +48). Power supply noise is 5 mV (Or -70 db re 15V @ 10 Mhz bandwidth). The circuit has just enough PSSR to get this down to 10's of uV in the output at full gain.
So with the 3 mV of RFI and power supply noise, it is not a very good preamp!
Adding an input filter and an RC to the PS will probably bring it up to acceptable.
No varactors were/are used to generate microwave frequencies at odd harmonics (usually) of a pump frequency. There was a 1mW at 120GHz one at ISSCC. The operation is not simple varying the capacitance but IIRC they snap on and off rather fast.
Unfortunately, a 'really high quality' all jfet input stage is very expensive today, due to the complexity of the circuit and very high cost of the ultra low noise jfets.
A transformer could be used, to get a balanced input with less complexity. I personally have not found a transformer at any price point that I consider as good no transformer at all, so I have elected to not use any in these very expensive phono stages. One having a retail price of about $19,000, the more expensive one probably priced at $60,000.
A transformer could be used, to get a balanced input with less complexity. I personally have not found a transformer at any price point that I consider as good no transformer at all, so I have elected to not use any in these very expensive phono stages. One having a retail price of about $19,000, the more expensive one probably priced at $60,000.
Yes, but typically the noise in question is at the lower frequencies. Either that or it's well above the audio band where it can be taken care of with simple filtering.
What, was this some sort of large home theater system where you're running 100 feet of cable to an active subwoofer or something?
How on earth are you getting the cable to pick up 60 volts of common-mode noise?
se
A sense of perspective has to brought into the balanced-single ended debate.
OF COURSE, when you are trying to move audio signals hundreds of feet, with parallel connections to other audio systems, in an environment that you don't know or can't control, THEN transformers can be a wonderful solution to ground loops, and common mode hum pickup. That is what it was originally designed for.
However if, like me, you have a phono player connected with maybe 4 feet or 1 meter of cable, it is just a 'waste of time' to have a balanced input. This is why I don't use balanced inputs in the CTC Blowtorch, and I don't have any problem.
However, IF I put the phono player in another room, let's say with 5 meters of cable, maybe a good balanced input or a transformer would be an ideal solution. This is WHY we put balanced in for the Constellation phono preamps. Rich guys sometimes actually do put their phono player in another room. But look at the COST! 4 times as many expensive jfets per channel. In my design, 45 per channel, count them! And you guys squawk about the cost of purchasing 2 low noise jfets per channel. '-) Could we reduce the number of jfets used? Yes, BUT at the cost of higher input noise and/or less dynamic range. For a 'best' phono preamplifier that we can design, we chose the all out, not compromised approach. If we didn't, someone else perhaps would, and we would not be the 'best' effort possible then. It is a design choice.
OF COURSE, when you are trying to move audio signals hundreds of feet, with parallel connections to other audio systems, in an environment that you don't know or can't control, THEN transformers can be a wonderful solution to ground loops, and common mode hum pickup. That is what it was originally designed for.
However if, like me, you have a phono player connected with maybe 4 feet or 1 meter of cable, it is just a 'waste of time' to have a balanced input. This is why I don't use balanced inputs in the CTC Blowtorch, and I don't have any problem.
However, IF I put the phono player in another room, let's say with 5 meters of cable, maybe a good balanced input or a transformer would be an ideal solution. This is WHY we put balanced in for the Constellation phono preamps. Rich guys sometimes actually do put their phono player in another room. But look at the COST! 4 times as many expensive jfets per channel. In my design, 45 per channel, count them! And you guys squawk about the cost of purchasing 2 low noise jfets per channel. '-) Could we reduce the number of jfets used? Yes, BUT at the cost of higher input noise and/or less dynamic range. For a 'best' phono preamplifier that we can design, we chose the all out, not compromised approach. If we didn't, someone else perhaps would, and we would not be the 'best' effort possible then. It is a design choice.
The target was a commercial product intended to be ina blister pack in about 2000 retail stores. It must always work or its not worth doing.
Example: One ungrounded AV receiver with a grounded subwoofer. Chassis leakage at the AV receiver sets its chassis at 1/2 line voltage. Very small current through the transformer to ground on the other side. Shields in transformer. Unbalanced cables ( consumer, RCA subwoofer output) Source impedance maybe several hundred Ohms, load impedance 20K. The leakage gets mixed with the audio on a high impedance transformer. A low Z transformer requires an active circuit to drive it which A) is too expensive a solution and B) requires power supplies and approvals. Once the circuit is active we did not need a transformer BUT at 5X markup its too expensive a product to have a customer base.
I heard some also place the balanced pre-pre, CD transport, balanced DAC, and balanced preamplifier in the adjoining chamber.
(could be my imagination playing havoc again)
Designing with expensive unobtanium parts is a JC trademark. Not looking for a modern solution and "sticking with what works" since the flower power times is another catch for the price-no-objection audio fashion market.
Since the target customers are those with deep pockets and a good chunck brown matter dangling in the deep space between the ears, complaining about "costs" is nothing but hypocrisy.
- Status
- Not open for further replies.