Re: Re: Re: Re: Re: Re: Re: ThermalTrak Design Considerations
What do you suspect that can go wrong with this approach. I am interested to know.
I would like to have a solution where there would be minimum interraction between adjusting Vbias on one end and adjusting the drift on the other for experimenting purpose
I believe that adjusting Vbias at room temperature for minimum distortion and then independently tempco at high temperature again for minimum distortion would be a very interesting experiment. Correlating this with sound quality is very interesting.
This is my opinion because it looks to me that the influence of temperature variations on sound quality is not well documented.
What do you think?
JPV
Bob Cordell said:
What do you suspect that can go wrong with this approach. I am interested to know.
I would like to have a solution where there would be minimum interraction between adjusting Vbias on one end and adjusting the drift on the other for experimenting purpose
I believe that adjusting Vbias at room temperature for minimum distortion and then independently tempco at high temperature again for minimum distortion would be a very interesting experiment. Correlating this with sound quality is very interesting.
This is my opinion because it looks to me that the influence of temperature variations on sound quality is not well documented.
What do you think?
JPV
I asked Douglas about Erik Margans amp
If you search forum " margan" theres a pdf from EWW circa 1987
If you search forum " margan" theres a pdf from EWW circa 1987
Re: I asked Douglas about Erik Margans amp
Ah yes, I remember it now.
I'm afraid I can't give you much of a reply though. I have respect for Erik Margan's abilities, but without replicating his tests that's about all I can say.
My own experience with circuitry designed to stop output devices turning off is that it does not decrease distortion, and can easily make it much worse.
Douglas
albin said:
Ah yes, I remember it now.
I'm afraid I can't give you much of a reply though. I have respect for Erik Margan's abilities, but without replicating his tests that's about all I can say.
My own experience with circuitry designed to stop output devices turning off is that it does not decrease distortion, and can easily make it much worse.
Douglas
Now I'm confused😕
Are the t-traks just a "bandaid" to compensate for an outdated
topology?? What advantage does near instant Tcomp have in
the real world?
Would not a well designed CFP with the Vbe/driver
die to die physical
connection not yield the same thermal stability/
quick response as the EF t-traks (I see wavebourns point).
If we are going to all the trouble of designing a precision
Vbias circuit to plug the trak's diodes into a CFP might be
better.
Re: I asked Douglas about Erik Margans amp
It is mentionned in this Douglas's page
http://www.dself.dsl.pipex.com/ampins/library/ampartew.htm
Margan published the idea previously in a short EWW "Circuit ideas".
At the time of publication, I was much more subjectivist than now and did not like the implementation I made of Margan's circuit in a JLH amp.
albin said:
It is mentionned in this Douglas's page
http://www.dself.dsl.pipex.com/ampins/library/ampartew.htm
Margan published the idea previously in a short EWW "Circuit ideas".
At the time of publication, I was much more subjectivist than now and did not like the implementation I made of Margan's circuit in a JLH amp.
We seem to have reached for the point where some real-life experimentation is needed. I'd better see if I can get hold of some samples...
A couple of thoughts occur.
The value of the VAS current-source will drift as it warms up, and may change with the supply-rail voltage. There is of course a well-known way to compensate for this by adding a resistor to a Vbe-multiplier.
A rather more challenging problem is that even if you have a perfectly stable bias generator, there is another influence on the quiescent conditions. If you have an unregulated power supply, and run the amplifier from a variac, turning it up and down varies the Vq. Last time I tried it on an EF Type 2 amplifier, changing the mains over the range 80% to 110% caused an 8 mV variation in Vq. This is nothing to do with the bias generator- it is caused by Early effect in the driver and output transistors. There is a section on this at the end of Chapter 13 in the Amplifier Handbook.
Not wishing to muddy the waters, but if we're aiming for accuracy it's one more thing to bear in mind.🙂
Douglas
A couple of thoughts occur.
The value of the VAS current-source will drift as it warms up, and may change with the supply-rail voltage. There is of course a well-known way to compensate for this by adding a resistor to a Vbe-multiplier.
A rather more challenging problem is that even if you have a perfectly stable bias generator, there is another influence on the quiescent conditions. If you have an unregulated power supply, and run the amplifier from a variac, turning it up and down varies the Vq. Last time I tried it on an EF Type 2 amplifier, changing the mains over the range 80% to 110% caused an 8 mV variation in Vq. This is nothing to do with the bias generator- it is caused by Early effect in the driver and output transistors. There is a section on this at the end of Chapter 13 in the Amplifier Handbook.
Not wishing to muddy the waters, but if we're aiming for accuracy it's one more thing to bear in mind.🙂
Douglas
DouglasSelf said:
I agree with you, we have to take care of all the influences if needed.
My initial thoughts are for an experiment : testing the influence on distortion and on aural quality of the tempco accuracy. With this circuit we can understand better the influence of transient temperature mistraking on distortion. We can also investigating the fact that we do need a different tracking to keep the condition gmR = 1 like I explained in the previous posts ( optimal Vq increases with temperature).
Measuring this and testing the aural value is interesting and could not be done easily without thermaltraks.
If the conclusion is that we need to track better then we should worry about the different effects you mentionned but then we know why.
JPV
G.Kleinschmidt said:
Hi Glen
Now that is an interesting notion. Splitting the VAS current-source in two is obviously easy; just use two current sources and connect them in parallel. The VAS transistor itself might be more of a challenge; how do you enforce current sharing there? Small emitter resistors?
Douglas
ostripper said:
My view is that the ThermalTraks are a breakthrough improvement to solving a historically difficult problem.
First of all, we do want the much faster thermal tracking speed of junction temperature. The slow tracking afforded by the traditional sensor on the heat sink allows a very large amount of thermal mis-tracking.
Secondly, the presence of the ThermalTrak sensing diode on the transistor die header significantly reduces the thermal attenuation that takes place between the transistor junction and the heat sink. Often, a 3 degree change in long-term junction temperature will only result in a 2-degree change in heat sink temperature under continuous transistor dissipation. This effect leads to serious mis-tracking when that transistor dissipation goes away.
The CFP is in principle more thermally stable because it is mainly the driver transistor Vbe that controls bias, and that Vbe changes much less with program material. However, the precision with which a CFP must be biased for optimum distortion and absence of gm doubling is much greater, so there is little net benefit in the end.
Cheers,
Bob
DouglasSelf said:
Hi Doug.
Yes. A typical unipolar VAS design may have a transistor acting as a current clamp to protect the VAS transistor, so the VAS transistor would be provided with an emitter degeneration resistor perhaps in the order of 22 ohms.
Suppose you add another VAS transistor in parallel, also with a 22 ohm emitter degeneration resistor. You now have a VAS with double the quiescent current, that is just as linear, but effectively with an 11-ohm emitter degeneration resistor (so, approximately, double the gain).
Cheers,
Glen
Well explained, fully understood. I'm not "anti-trak" but the only
sources I could find sell bulk lots.. so it is an issue
of availibility.. I.E. I'm jealous..😀
Os
For you who are in the US you can buy the thermaltraks in small quantities from digikey.com and arrownac.com. Arrow seems to be selling out their last non-lead-free parts so they are very cheap there 😀
I guess the low price of $0.57 and $0.78 is because most manufacturers want to be able to sell to the EU and can't use them because their lead-free solder baths would get contaminated... :O
Good for us though 🙂
I bought some, the shipping was pretty expensive overseas though, about 50€ IIRC. I could sell a couple if anyone in the EU wants a few.
Good for us though 🙂
I bought some, the shipping was pretty expensive overseas though, about 50€ IIRC. I could sell a couple if anyone in the EU wants a few.
A group buy on these via a member across the pond might elicit much interest, do you not think?
Regards,
Brian.
Regards,
Brian.
- Status
- Not open for further replies.