Not sure what you mean by "6 dB improvement." Seems like you're thinking of balanced in terms of line level interfaces, which isn't quite the same thing.
True.
se
In 'practice' it is not very high. This is the often seen difference in this forum... theory vs practice. In order to achieve the full potential of a balanced transformer's ability to reject noise, the Z on each leg would need to be equal. In practice, 30dB would be a very good figure, in most cases. Easily much less.
The power is balanced by the 230v feed to the USA home but in the home, most appliances (audio/video etc) are run on one side or the other... unbalanced. It would work well to have all the A-V gear powered by 230vac if you were designing the home wiring.... as some studio's do here in USA. But, you still run into the degree of balance for rejection.
An isolation transformer works better in most cases.
THx-RNMarsh
Last edited:
I ordered the completed one..... But, I think I'll get more use with it in audio.... I dont have T&M ground issues if I unplug the T&M's portable computers' battery charger when taking a measurement. I also use ground isolation transformer on other connected test equipment.
THx-RNMarsh
balanced power
Of course there are these folks: Products Index
I may have told the story about a defective powered loudspeaker that found its way into Keith Johnson's hands for evaluation. I was still at Harman at the time and was told "someone" had reported a defect in a sample of the particular system. The person telling me about this told me he was sure it was nothing about which to be concerned, and it was probably some customer who had decided to modify the unit and screwed up, and that this sort of thing had happened at his former job with LabTech. Not to worry.
I got a copy of the email exchange and must have made a loud noise while reading it. "Do you know who this is??" He did not. I informed him.
Keith had embraced the Equitech balanced power approach and had his lab appointed thus. Unfortunately, the defect in the speaker system had shorted the AC neutral to the system common. Although Keith had destroyed the smoking gun evidence by disassembling the unit (which we had made fairly difficult to do), he believed that we had designed it intentionally to have neutral connected thus---which would have been gross negligence and outright insanity.
When hooked up to his Pacific Microsonics DAC it immediately destroyed its output stage and who knows what else.
The best we could imagine was that a bracket insert in the plastic had been misinstalled and crunched down on some portion of one of the PCB assemblies and effected the short circuit. An inspection procedure was implemented to check for proper bracket position. As far as I know KOJ was the only person who had the problem, (!) which for other users would have resulted in excessive hum and noise but probably no destruction---unless of course the mains polarity happened to be reversed, which of course would be very dangerous.
I spoke to Keith and assured him that this was anomalous, and a massive recall was averted.
Of course there are these folks: Products Index
I may have told the story about a defective powered loudspeaker that found its way into Keith Johnson's hands for evaluation. I was still at Harman at the time and was told "someone" had reported a defect in a sample of the particular system. The person telling me about this told me he was sure it was nothing about which to be concerned, and it was probably some customer who had decided to modify the unit and screwed up, and that this sort of thing had happened at his former job with LabTech. Not to worry.
I got a copy of the email exchange and must have made a loud noise while reading it. "Do you know who this is??" He did not. I informed him.
Keith had embraced the Equitech balanced power approach and had his lab appointed thus. Unfortunately, the defect in the speaker system had shorted the AC neutral to the system common. Although Keith had destroyed the smoking gun evidence by disassembling the unit (which we had made fairly difficult to do), he believed that we had designed it intentionally to have neutral connected thus---which would have been gross negligence and outright insanity.
When hooked up to his Pacific Microsonics DAC it immediately destroyed its output stage and who knows what else.
The best we could imagine was that a bracket insert in the plastic had been misinstalled and crunched down on some portion of one of the PCB assemblies and effected the short circuit. An inspection procedure was implemented to check for proper bracket position. As far as I know KOJ was the only person who had the problem, (!) which for other users would have resulted in excessive hum and noise but probably no destruction---unless of course the mains polarity happened to be reversed, which of course would be very dangerous.
I spoke to Keith and assured him that this was anomalous, and a massive recall was averted.
This side of the pond we use 4 wire three phase, so a power imbalance on the legs behaves slightly differently than the 3 wire three phase you guys use, a power imbalance leads to a voltage between neutral and ground, your system leads to phase errors for any three phase loads.
I have met and talked face-to-face with the inventor and owner of EQuiTech extensively about the balanced transformer approach and I had measured its performance. he found, as I did also, it doesnt reject very much due to load imbalances, as i said. A good idea on paper but not so good in practice.
THx-RNMarsh
Last edited:
The main benefit of "balanced" in this sense isn't the transformer's rejection of noise on the AC mains. The benefit is in the secondary being balanced. Which I don't think you quite understand.
se
Dan, subjectively is this now the best it's ever been, or has it merely been restored to a level you've experienced before? If a genuine improvement over previous listening, in what area have gains been made?
Back to the line amp?
The question was posed as to the effect of the source resistors in the balanced complementary topology shown by JC in post 68108. Besides affording a place to adjust d.c. balance, R1-R4 also provide some local feedback before the application of global feedback. This is a fairly easy situation to analyze, especially if the load in a given drain of any input device is fairly low Z, as open loop we have about all of the transconductance, gm, in action, and thus a voltage gain at a given JFET source given by Av = Rs/(1/gm + Rs). The gate-source capacitance is therefore effectively reduced based on how close to unity gain we get. Neglecting the reverse transfer capacitance, the effective input C at low-mid frequencies is Ceff = (1 - Av)Cgs. Any variations in that device capacitance, that will amount to distortion mechanisms based on the input driving impedance, are correspondingly alleviated.
But when there is sufficient loop gain, its effect is to make the JFET sources closely follow the gates---that is, the device Cgs's are nearly zero since the voltage gains are close to unity. So, that part of the input distortion is made small. This comes at the expense of higher input noise from the thermal noise of the resistors. But there is the additional advantage of making the equivalent transconductance more constant with signal swing.
The drain loads are not quite that low however. Depending on the details, the bipolars do load the 1k pullup/pulldown resistors. For one set of models and with relatively low closed loop gain, I see a Miller multiplier at each drain of 1 + 5.34, that is, times 6.34, due to the local voltage gain of -5.34.
This is for some garden-variety bipolars and my models of the 2SJ74 and 2SK170, and with 100 ohms for each of R1 through R4, Rfeedback = 1k, and the specified 1k output loads.
By simulating a rather high impedance drive, and operating the circuit open loop (with each junction of R1-R2 and R3-R4 tied to common, or to a small bias voltage if necessary to zero the outputs) and looking at the frequency response at the inputs, one can deduce the input capacitances. I see about 134pF in this case.
Closing the loop reduces these to about 15.3pF for low-mid frequencies. So this corresponds to virtually all of the input C being due to the reverse transfer capacitances and their Miller multipliers. Since those capacitances vary with voltage, whether it's a linear or nonlinear variation, they will constitute a distortion mechanism for distortion at the input. When the driving signal is low impedance the effects will be small.
There is a partially compensatory effect due to the complementary arrangements on each side of P and N, as with voltage swing one capacitance is changing in opposite polarity to the other. If the devices were truly complementary this effect would provide ideal cancellation---but the P capacitances are higher than the N's.
As for total distortion performance and its spectrum, that's helped considerably by the balanced configuration and the predominant second-harmonic-dominated behavior of the JFETs. Thus with good device matching good cancellation can be achieved.
The question was posed as to the effect of the source resistors in the balanced complementary topology shown by JC in post 68108. Besides affording a place to adjust d.c. balance, R1-R4 also provide some local feedback before the application of global feedback. This is a fairly easy situation to analyze, especially if the load in a given drain of any input device is fairly low Z, as open loop we have about all of the transconductance, gm, in action, and thus a voltage gain at a given JFET source given by Av = Rs/(1/gm + Rs). The gate-source capacitance is therefore effectively reduced based on how close to unity gain we get. Neglecting the reverse transfer capacitance, the effective input C at low-mid frequencies is Ceff = (1 - Av)Cgs. Any variations in that device capacitance, that will amount to distortion mechanisms based on the input driving impedance, are correspondingly alleviated.
But when there is sufficient loop gain, its effect is to make the JFET sources closely follow the gates---that is, the device Cgs's are nearly zero since the voltage gains are close to unity. So, that part of the input distortion is made small. This comes at the expense of higher input noise from the thermal noise of the resistors. But there is the additional advantage of making the equivalent transconductance more constant with signal swing.
The drain loads are not quite that low however. Depending on the details, the bipolars do load the 1k pullup/pulldown resistors. For one set of models and with relatively low closed loop gain, I see a Miller multiplier at each drain of 1 + 5.34, that is, times 6.34, due to the local voltage gain of -5.34.
This is for some garden-variety bipolars and my models of the 2SJ74 and 2SK170, and with 100 ohms for each of R1 through R4, Rfeedback = 1k, and the specified 1k output loads.
By simulating a rather high impedance drive, and operating the circuit open loop (with each junction of R1-R2 and R3-R4 tied to common, or to a small bias voltage if necessary to zero the outputs) and looking at the frequency response at the inputs, one can deduce the input capacitances. I see about 134pF in this case.
Closing the loop reduces these to about 15.3pF for low-mid frequencies. So this corresponds to virtually all of the input C being due to the reverse transfer capacitances and their Miller multipliers. Since those capacitances vary with voltage, whether it's a linear or nonlinear variation, they will constitute a distortion mechanism for distortion at the input. When the driving signal is low impedance the effects will be small.
There is a partially compensatory effect due to the complementary arrangements on each side of P and N, as with voltage swing one capacitance is changing in opposite polarity to the other. If the devices were truly complementary this effect would provide ideal cancellation---but the P capacitances are higher than the N's.
As for total distortion performance and its spectrum, that's helped considerably by the balanced configuration and the predominant second-harmonic-dominated behavior of the JFETs. Thus with good device matching good cancellation can be achieved.
Steve, we get it.
the main advantage I have seen when an Equi-tech system has been installed is that the whole system has been gone thru and renewed. No more loose connections, questionable sockets, etc. and it has an isolation transformer built in, it doesn't matter if its single ended out or balanced out, the bringing the system up to snuff and iso'd is what matters.
You get most (all but the last 6 db max in theory) without the expense of a highly marked up product.
Its just the facts, oh well.
Alan
the main advantage I have seen when an Equi-tech system has been installed is that the whole system has been gone thru and renewed. No more loose connections, questionable sockets, etc. and it has an isolation transformer built in, it doesn't matter if its single ended out or balanced out, the bringing the system up to snuff and iso'd is what matters.
You get most (all but the last 6 db max in theory) without the expense of a highly marked up product.
Its just the facts, oh well.
Alan
Actually some of this is a bit off. Because, closed-loop, the motion at each drain is reduced a lot, the Miller multiplier argument is a significant overestimate of the Crss contribution to input capacitance. What is the primary contribution: just the input voltage swing. Most of the Miller multiplication is gone with the loop closed. This agrees fairly well with other determinations of Crss for the two types of JFET.
Last edited:
are there any Engineers in the house - or just these useless blowhards?
or anyone who can Google, read
er no - the device doesn't do USB 2.0 High Speed 480 Mb - no one does with these single chip serial isolators so far as I know
USB Low Speed 1.5 Mb, Full Speed 12 Mb /= High speed 480 Mb
and you need to know if you're using USB Audio Class 1 or 2
The Well-Tempered Computer
this very recent page shows fiber optic USB bridges as the only current High Speed 480 Mb isolation
The Well-Tempered Computer
or anyone who can Google, read
er no - the device doesn't do USB 2.0 High Speed 480 Mb - no one does with these single chip serial isolators so far as I know
USB Low Speed 1.5 Mb, Full Speed 12 Mb /= High speed 480 Mb
and you need to know if you're using USB Audio Class 1 or 2
The Well-Tempered Computer
this very recent page shows fiber optic USB bridges as the only current High Speed 480 Mb isolation
The Well-Tempered Computer
Last edited:
Who's "we"?
Still don't know where this "6 dB max in theory" is coming from. 6 dB max what, and what theory?
What facts? I've just seen a 6 dB figure and some gibberish.
Can someone who knows waltzingbear better than I do tell me if I'm just wasting my time here? Will I just be getting Richard flashbacks?
se
- Status
- Not open for further replies.