Hi Joshua!
This discussion not neccesary. As you are going to use them as the top part of a cascode, with FETs at the bottom, the FETs are gonna decide the distortion. The tube will only have a secondary effect.
Thorsten
This discussion not neccesary. As you are going to use them as the top part of a cascode, with FETs at the bottom, the FETs are gonna decide the distortion. The tube will only have a secondary effect.
Thorsten
At the expected swing (maybe 100mV max), you'll have to really screw things up to see anything measurable above the second.
Indeed.
On top of that, calculated HD from real curves aren't the same as measured HD.
Please let me state again:
I'm interested in measured results (and, if possible, also listening impressions). More than THD figures, I'm interested in the higher harmonics HD figures.
I'm not interested in calculated results.
Whether I'm right or wrong here – this is what I'm interested in.
Thanks.
That is correct, at least partially. However not all tubes sound the same, even when cascoding JFET's. Else, it wouldn't matter what tubes will be used. I don't think that the choice of tubes doesn't matter.
It applies to any component with smooth characteristics and sufficiently small signals. 'Smooth' means fully differentiable and so a Taylor series can be used to model it. Then any sufficiently small signal will give negligibly small high order distortion. As I said, mathematics is on your side.
As Dave says, pretty much, yes. Perhaps there's some remote cutoff tube where this isn't true at these levels, but that would even be surprising. I've confirmed this with at least a dozen different tube types. With the gain topology that Allen used (not specific to him, of course), the distortion signature is totally dominated by the second stage. In my phono stage, I can easily see 3:1 distortion differences between different ECC88s in the second hole.
Again, the important parameters for the first stage tube are noise, microphonics, and gain. Low plate resistance is important in grounded cathode stages, but irrelevant in cascodes.
Thanks, Dave and Stuart.
I'll go for C3g.
Like in Allen's RTP3C phono stage, there will be only one stage (for MM/MC, without MC step-up transformer), differential, or LTP pair of JFET's (for each channel), each FET cascoded by a tube (diode strapped C3g).
Like in Lynn's Raven preamp, the tubes will be loaded by a trafo.
I'll have to make, first some simulation and then some measurements, in order to see if and how I may also implement Lynn's Ultrapath configuration.
Another possible headache may be calculating and implementing the passive RIIA equalization, however at this stage this is the least of my concerns.
I'll go for C3g.
Like in Allen's RTP3C phono stage, there will be only one stage (for MM/MC, without MC step-up transformer), differential, or LTP pair of JFET's (for each channel), each FET cascoded by a tube (diode strapped C3g).
Like in Lynn's Raven preamp, the tubes will be loaded by a trafo.
I'll have to make, first some simulation and then some measurements, in order to see if and how I may also implement Lynn's Ultrapath configuration.
Another possible headache may be calculating and implementing the passive RIIA equalization, however at this stage this is the least of my concerns.
When using a transformer load, don't lose sight of the fact that the driving impedance will be quite high. That's normally not a positive.
Would you care to elaborate?
(From what I know so far, the higher the load to the tubes, the better, or am I wrong here? Or did you mean the output impedance of that stage?).
The loading transformer can be 1:1, or 1:x, or x:1.
On top of that, this phono stage will be built along a preamp, in the same casing. So it's up to me to choose the value of the potentiometer at the input of the preamp (within reasonable limits).
Should it be necessary to add a second stage – I prefer it over one stage, resistor loaded, capacitor coupled. (The preference here isn't driven by mere technical, or engineering, considerations).
The output impedance of the source driving the transformer (i.e., the plate of the tube). Cascoding makes the effective plate resistance very large (remember, cascode I-V plate curves look like a pentode).
This is the hazard in trying to design by cutting and pasting- IMO, you ought to step back and think about the overall engineering in light of what you want the preamp to do, as opposed to patching together disparate design elements from well-engineered designs that have little in common.
This is the hazard in trying to design by cutting and pasting- IMO, you ought to step back and think about the overall engineering in light of what you want the preamp to do, as opposed to patching together disparate design elements from well-engineered designs that have little in common.
Thank you.
I will rethink it, after making some calculations and simulations.
At this stage, what I do like to maintain, if in accord with engineering considerations, is:
1. Balanced (and differential) from input of the phono stage to the output of the preamp. (That is, apart from the output of the preamp, which will be both balanced and SE, and apart from the inputs of the preamp, which will be some SE and some balanced, via switchable input transformer to the preamp).
2. No step-up transformer for MC cartridge.
3. Switchable MM/MC phono cartridges (I have one of each. When the MC is being shipped for re-tipping, I use the MM).
4. Only transformers (no capacitors) in signal path (other than in the RIIA equalization).
5. Of course (it goes without saying), very good measured technical aspects, like THD, SNR, good immunity to RFI and flat, wide-band, frequency response.
That being said, on a first glance, as for output impedance of the phono stage, I see no marked difference between resistor loading vs. transformer loading. Yet again, I'll reconsider it.
I should add to the technical considerations ample headroom and instantaneous recovery from overload.
Also, I should consider the desired Ip of the chosen tubes (for best linearity) vs, the IDSS of the available JFET's (or possibly BJT's).
Also, I should consider the desired Ip of the chosen tubes (for best linearity) vs, the IDSS of the available JFET's (or possibly BJT's).
You need to get the architecture right before thinking too much about details. So far, you just have a wish-list; reality may force compromises on you.
I would start from the RTP3C and use the 6992. This valve can be quite linear. I don't thinK D3a, G3g etc. can do better and actually can be quite tricky. At some point, somewhere between 0.03% and 0.1% there is irreducible distortion and without feedback or other useful technique it is impossible to go lower. The only difference between the two group of valves above is that more likely that small residual amount will be mostly 2nd harmonic for the 6922 and mostly 3rd harmonic for the triode strapped pentodes. It's constitutional! Having more electrodes perfect "symmetry" is more difficult. Cancelling small amounts of 2nd harmonics with two cascaded stages is possible without compromising the other things.
As others said you have other priorities in a phono amp than linear behavior with large signals.
The Ale Moglia pages are very useful. There are quite a lot of real THD measurments and in some cases the results are quite representative having measured a good number for each type.
If you want to know precisely just send him your specific velves and get them tested!
Valve/Tube Testing Service | Bartola Valves
Last edited:
That turns out not to be the case. Triode-strapped D3a shows much lower 3rd than ECC88 types (including 6922) at swings large enough to be able to see this (not relevant for first stage of phono). The only triodes that come close to the triode-strapped pentode are ECC83, which are very tricky to use properly. I haven't measured C3g.
Hi,
Thank you.
I described the architecture I have in mind: one stage MM/MC, differential JFET's cascoded by triodes (or triode-strapped pentodes), loaded by inter-stage transformer, passive RIIA equalization, in one case with the preamp.
What are your comments about the above architecture?
- Status
- Not open for further replies.