|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
81.216.133.244
In Reply to: RE: Modified Futterman OTL amplifier posted by cellai@space.it on June 26, 2008 at 11:45:57
In general Spice simulation of distortion is difficult and with 6C33C it is even more difficult. The main problem is to find a good enough model that resembles the real characteristics of the tube, the 6C33C exhibit some unusual behavior with a kink in the Ia characteristics at low currents and this is difficult to simulate in a model. I have used all available models for spice simulation of the 6C33C that I have found and no one give results that are accurate for distortion, the best I have found so far is from here http://www.excem.fr/download/usergui5.pdf
It is difficult to see from the schematics that you have posted but as far as I can see you have not included any balance compensation in the driver stage. As I wrote earlier a SEPP is unbalanced and need a special driver arrangement for compensation, the most well known is the Futterman. In the Futterman with split load phase inverter the cathode resistor is returned to the output rather than ground and this will compensate for the unbalance, but I can' see that you are using that circuit, (sorry but the resolution of the scematics is not good enough to see clearly).
In my own OTL where I use the Futterman connection the 2nd order product are 55dB down before feedback, I haven't removed the Futtterman connection and measured but it would certainly be much worse.
Regards Hans
Follow Ups:
I am wondering if it is possible to partially compensate for the intrinsic gain imbalance of the output stage (with no feedback and without resorting to a Futterman phase splitter returned to output), at least in the case of a load equal to the nominal speaker impedance (8 Ohm).
The 6080 (double triode) tubes that I have used have a plate resistance of 300 Ohm and mu=2 (gm = 6.5 mA/V). The parallel of four 6080 tubes exhibits a nominal plate resistance (rp) of 300/4=75 Ohm.
Based on these values and standard formulas the nominal gain of the upper triodes is 0.1616 = mu/((mu+1)+rp/8 Ohm) and the nominal gain of the lower triodes is 0.1927 = mu*8 Ohm/(rp+8 Ohm).
In other words the gain imbalance is 19.2% (0.1927/0.1616).
If one fixed 6 Ohm resistor were added in series with the cathodes of the lower triodes the corresponding gain would go down from 0.1927 to 0.1584 = mu*8 Ohm/(rp+8 Ohm+(1+mu)*6 Ohm) and the resulting nominal imbalance would be less than 3%.
Could it be an alternative (for even harmonic reduction) worth to try?
Yes, it would be possible to compensate for the gain difference but the output impedance from each tube would still be different meaning that the balance would be lost if the load impedance would vary as it does in a real speaker. The futterman connection, (and some other methods) compensate for the gain and the output impedance difference and the balance is therefore maintaned also when the load impedance vary.
I have feeling that you don't like the Futterman compensation, do you want to explain why?, (I am just curious).
There has been many papers written about methods for how to compensate the unbalance of the SEPP amplifier I can recommend the paper by Hiroya “Analyses of Drivers for Single-Ended Push-Pull Stage,” Hiroshi Amemiya, IRE Transactions-Audio, Sept-Oct 1955" that is also mentioned by John Broskie
Regards Hans
I have provisionally tried a 5.4 Ohm (in series with the cathodes of the SEPP lower triodes) for fixed compensation of the SEPP nominal gain imbalance (0% nominal gain imbalance @8 Ohm).
This simple modification was mainly aimed at assessing if I was able to listen to some audible difference (with and without such a fixed balance compensation).
The limited listening test performed so far has been clearly based on impressions 'prior and after' the modification (hence not by a live comparison of the switched outputs of two simultaneously available amplifiers), as the OTL amplifier HW had to be always the same.
Theoretically the calculated 2nd harmonic (closed loop) distortion should have gone down from a supposed level of something less than -30 dB (no compensating resistor) to something better than -40 dB (with compensating resistor).
I must say that if there is a difference (and I believe there must be) it can be only ascertained by some direct measurement and spectral analisys (not available at this time).
By my ears only (over some hours of listening to my favourite classic music) I can neither say if there is a significant difference nor if the sound is subjectively better (sorry, probably I am not a 'golden ear').
This is the reason why I am not going to modify further the present OTL design.
A fully automatic SEPP gain imbalance compensation will be the matter of my next OTL (likely a true 'modified Futterman' like yours).
For sake of completeness the calculated compensating resistor for 16 Ohm speakers is 10.65 Ohm.
A spreadsheet is available to potentially interested DIY readers for calculation of compensating resistor (send a request to cellai@space.it).
Best Regards
Luca
I have nothing against Futterman....rather the contrary is true!
In the present context I would just prefer to compensate for the intrinsic imbalance of the output stage (I remind that the OTL amplifier is in use since a month ago and that I am quite satisfied of its current performance) by adding a simple resistor (rather than by adding a new 450-500 V rail and by introducing some strong modification of the current OTL schematic).
An Italian proverb (a sort of Murphy's law) states that when something is almost good enough, you should not modify it, otherwise it fails.
The following table shows the calculated variation of the gain of the upper and lower triodes of the output stage as a function of the speaker impedance (from 1 Ohm to 17 Ohm, assuming a fixed compensation resistor of 5.5 Ohm, that is, four 22 Ohm resistors in parallel).
The imbalance with no compensating resistor at all is 19.8%.
With a 5.5 Ohm compensating resistor (that is best for 8 Ohm speakers) the imbalance is less than one half (i.e. approximately 10%) when the speaker impedance varies from 4 to 14 Ohm. A good result for a simple resistor, don't you agree?
Tube Type 6080 (4X) mu=2, rp=75
Speaker Up Lower Up-Low/Up
1 0.0216 0.0256 -0.157
2 0.0428 0.0494 -0.134
3 0.0635 0.0714 -0.111
4 0.0838 0.0920 -0.089
5 0.1036 0.1111 -0.067
6 0.1231 0.1290 -0.046
7 0.1421 0.1458 -0.025
8 0.1608 0.1616 -0.005
9 0.1791 0.1765 0.015
10 0.1970 0.1905 0.034
11 0.2146 0.2037 0.054
12 0.2319 0.2162 0.072
13 0.2488 0.2281 0.091
14 0.2654 0.2393 0.109
15 0.2817 0.2500 0.127
16 0.2977 0.2602 0.144
17 0.3134 0.2698 0.161
Last but not least I have also found that the sensitivity of the % of residual imbalance does not vary significantly with mu and rp: this is really intriguing.
Good result but maybe not good enough, 10% imbalance is more or less the same as 10% 2nd order distortion or -20dB, in a push pull stage I would look for at least -40dB or 1% which then calls for less than 1% unbalance for all loads.
As I described earlier my own amp has -55dB with 8 ohm and about -50dB with 4 ohm but then I also tune the phase splitter by adjusting one of the resistors, as I don't use matched output tubes it is necessary, without tuning the phase splitter I easily get -40dB or 1% 2nd order distortion.
Regards Hans
FAQ |
Post a Message! |
Forgot Password? |
|
||||||||||||||
|
This post is made possible by the generous support of people like you and our sponsors: