Joachim,
I think you have made some excellent points in all these posts. WRT INA
topology, I am assuming you mean a fully symmetrical instrumentation
topology. If you refer to Samuel Groners excellent opamp measurement pdf
it is clear that common mode distortion is a major player in opamp non
linearity and especially at higher frequencies. Using the 'INA' topology will
largely cancel this form of distortion.
WRT measurements of these pre amps , I would again do a 20kHz
measurement where CM distortions will show much clearer, a THD
versus frequency measurement and note how much it all tilts up toward
the HF. Note my post on power amp measurements.
cheers
Terry
Terry, i did that. You can see my MPP tread. Distortion in a well designed Opamp stage, be it harmonic, be it intermodulation, be it Belcher or whatever fancy test you can throw on them, is a non issue provided you do not drive it into a too demanding load and keep the voltage level resonable. My experience tells me that "magical sound" comes from carefully designed noise and distortion at a low level. Too high distortion then sounds foggy and uninformative unless a musician just wants that on the recording side.
John
There is a way out and that is to build a direct coupled valve preamplifier if you are keen , you just need of coarse some protection on the output ( I prefer a relay here).
Its surprising what they sound like without all those coupling caps.
I think you could also do an RIAA stage that could be direct coupled all the way through too.
Regards
Arthur
Hello Joachim
What do you imagine a low distortion valve stage to sound like do you think it would sound like a low THD silicon stage. In both cases I am talking about designs with no coupling capacitors.
I think ones the distortion is under -80dB, low order, exponentially falling, noise is not audible and frequency response is better then 200kHz -3db, tubes and transistors can sound surprisingly similar. Parts choice matters in silicon circuits too in my experience.
A tube poweramp with transformer output and low damping factor will always sound different from a good solid state design. In case the tube amp sounds better i asume it has some artfully designed distortion that makes the music sound better. I have no problem with that as long as i can enyoy the result and fatique does not set in.
I was told by David Salz of Wire World that there exists a tube manufacturer that sells to the State of America tube amplifiers that are extremely fast and have very low distortion, like silicon based circuits, so it can be done. Hewlett - Packard and Tectronics ( for that matter also Rhode & Schwarz and B&K ) must have known too how to design like that in the old days.
A tube poweramp with transformer output and low damping factor will always sound different from a good solid state design. In case the tube amp sounds better i asume it has some artfully designed distortion that makes the music sound better. I have no problem with that as long as i can enyoy the result and fatique does not set in.
I was told by David Salz of Wire World that there exists a tube manufacturer that sells to the State of America tube amplifiers that are extremely fast and have very low distortion, like silicon based circuits, so it can be done. Hewlett - Packard and Tectronics ( for that matter also Rhode & Schwarz and B&K ) must have known too how to design like that in the old days.
For better understanding of the advantages and tradeoffs discussed here in input topologies, I might elaborate a little:
INAMPS are the abbreviation of Instrumentation Amps, and have been used for many decades. They are essentially high common mode rejection balanced inputs that are relatively unaffected by source impedance and other factors. 120dB or 1 million to one is possible at low frequencies, and various discrete versions and IC versions have been made over the decades. One common type is the so called 3 amp configuration, composed of 3 op amp type designs put together to make 1 gain stage. This is apparently the type of gain stage mentioned recently here by Joachim, and is the usual front end of our distortion analyzers, from HP, ST, or AP.
This type of stage is relatively 'bullet proof', but it has the problem of being about 6 dB noisier than the quietest discrete or IC configuration, due to the fact that 4 devices are in SERIES at the input. This is what makes it twice as noisy as what is possible.
When working with MC cartridges, this can be a problem, as their output level is so low.
INAMPS are the abbreviation of Instrumentation Amps, and have been used for many decades. They are essentially high common mode rejection balanced inputs that are relatively unaffected by source impedance and other factors. 120dB or 1 million to one is possible at low frequencies, and various discrete versions and IC versions have been made over the decades. One common type is the so called 3 amp configuration, composed of 3 op amp type designs put together to make 1 gain stage. This is apparently the type of gain stage mentioned recently here by Joachim, and is the usual front end of our distortion analyzers, from HP, ST, or AP.
This type of stage is relatively 'bullet proof', but it has the problem of being about 6 dB noisier than the quietest discrete or IC configuration, due to the fact that 4 devices are in SERIES at the input. This is what makes it twice as noisy as what is possible.
When working with MC cartridges, this can be a problem, as their output level is so low.
This leads to what is the BEST way to make an input stage for MC cartridges. At the moment, I am designing two separate phono preamps with different inputs. One is single ended, the other is balanced. Neither is the traditional 3 op amp type design.
This is in an effort to keep noise to a minimum, as the 3 op amp approach is the noisiest option.
With transformerless design, the quietest approach is with single ended input, because the active device that are in SERIES are minimized. In fact, for the SAME number of input devices, you can get a 6dB improvement in S/N, by paralleling the previous series devices. This is the approach used in the Vendetta Research phono preamp, and recently in the JC-3 phono stage.
However, the trend, these days is balanced. This usually means phono input cables that are balanced as well, and RCA connectors are replaced with XLR. This is the approach used in the new Constellation phono preamp that I am now designing, and I presume, the Sequerra phono stage.
In fact, if you are just using an RCA connector with a traditional shielded phono cable, then transformerless balanced input should be next to useless. However, with a balanced phono cable, a certain resistance to picking up hum and external interference is to be expected.
Transformers are yet another approach, AND with a virtually perfect transformer, you can do better than any other way, BUT finding and AFFORDING a nearly perfect transformer is not so easy. Only in the most extreme cases, is a transformer absolutely necessary, and its rated distortion may not be as important as to how it reacts when ultra low frequency warp frequencies are added to the mix. This was the primary problem with transformers from earlier times, like 35 years ago, and why we strove to eliminate them with transformerless inputs.
This is in an effort to keep noise to a minimum, as the 3 op amp approach is the noisiest option.
With transformerless design, the quietest approach is with single ended input, because the active device that are in SERIES are minimized. In fact, for the SAME number of input devices, you can get a 6dB improvement in S/N, by paralleling the previous series devices. This is the approach used in the Vendetta Research phono preamp, and recently in the JC-3 phono stage.
However, the trend, these days is balanced. This usually means phono input cables that are balanced as well, and RCA connectors are replaced with XLR. This is the approach used in the new Constellation phono preamp that I am now designing, and I presume, the Sequerra phono stage.
In fact, if you are just using an RCA connector with a traditional shielded phono cable, then transformerless balanced input should be next to useless. However, with a balanced phono cable, a certain resistance to picking up hum and external interference is to be expected.
Transformers are yet another approach, AND with a virtually perfect transformer, you can do better than any other way, BUT finding and AFFORDING a nearly perfect transformer is not so easy. Only in the most extreme cases, is a transformer absolutely necessary, and its rated distortion may not be as important as to how it reacts when ultra low frequency warp frequencies are added to the mix. This was the primary problem with transformers from earlier times, like 35 years ago, and why we strove to eliminate them with transformerless inputs.
If you run across a "useless" Cotter transformer at a good price, let me know.
Today, I work very closely with Mitch Cotter. I just heard from him, this last weekend. I must ask him about that specific transformer. My measurements that I did about 30 years ago, showed me that it has an effective self noise of about 10 ohms or .4nV/rt Hz. Today we can get 5 ohm equivalent noise, or so, or hopefully .3nV/rt Hz. Expensive, though.
- Status
- Not open for further replies.