Personally, I wouldn't have thought that it could be stabilized well enough to even be accommodated by some offset in the plate voltages, but MJ has built it and it works after some fiddling with the CCS set points. I don't know for sure if he temperature compensated the CCS or had them thermally linked or whatever.
Balance between the two plate CCS can be taken up a bit by the "compliance" of a triode's lowish plate resistance (i.e., the tube's variation in plate current with plate voltage), but I am also surprised that he got stability with a tail CCS.
Balance between the two plate CCS can be taken up a bit by the "compliance" of a triode's lowish plate resistance (i.e., the tube's variation in plate current with plate voltage), but I am also surprised that he got stability with a tail CCS.
I don't have the book but, from the way you describe it, I would be confused too! The conflict between the plate load CCSs and the tail CCS is an obvious question to raise. 
Also, AFAIK, the whole idea of a CCS in the tail of an LTP is to achieve 'perfect' AC balance between the two halves, provided that the plate loads are of precisely equal resistance. I don't see how that can be guaranteed with CCS loads.
Ordinarily, a CCS load for a triode is useful in giving an almost horizontal load line, meaning maximum gain (almost the same as mu) and minimum distortion. A solid-state CCS can also mean only a small drop between B+ and the triode plate. These are attractive benefits, but CCS plate loads seem to be out of place here.
I'd love somebody to explain why I'm wrong!
Also, AFAIK, the whole idea of a CCS in the tail of an LTP is to achieve 'perfect' AC balance between the two halves, provided that the plate loads are of precisely equal resistance. I don't see how that can be guaranteed with CCS loads.
Ordinarily, a CCS load for a triode is useful in giving an almost horizontal load line, meaning maximum gain (almost the same as mu) and minimum distortion. A solid-state CCS can also mean only a small drop between B+ and the triode plate. These are attractive benefits, but CCS plate loads seem to be out of place here.
I'd love somebody to explain why I'm wrong!
That circuit caught my attention when I read Morgan’s fine book too. His caption reads: “Fine adjustment of tail current sets anode voltages”. The anode loads are fairly simple PNP BJT collectors with bases biased by a zener and a diode drop from the B+. These loads would not have very high resistances compared to many “CCS” designs. In other words, this circuit is stable/adjustable only to the extent that the current sources are not really constant. If these loads were cascoded or used some other extremely high resistance CCS design, the “fine adjustment” would be probably become “impossibly touchy”. I don’t think Morgan was proposing this particular circuit as a finished audio stage, but rather as a teaching example.
Guys,
A simple approach is here: http://www.diytube.com/st70/diytube_st70_B.pdf
I can assure you this approach sounds extremely good and works very well. Simple too.
Regards,
A simple approach is here: http://www.diytube.com/st70/diytube_st70_B.pdf
I can assure you this approach sounds extremely good and works very well. Simple too.
Regards,
Hi Brian
This explanation sounds good, but I have not found it in the third edition. Also, it does not sound as if the schematic on the third edition is a teaching example, a thing he mentions explicitly for other examples, as the DC coupled headphone amp on page 195.
SY, thanks for enquiring Morgan Jones directly. The man must not be happy you got his book, and most important, his e-mail
Erik
This explanation sounds good, but I have not found it in the third edition. Also, it does not sound as if the schematic on the third edition is a teaching example, a thing he mentions explicitly for other examples, as the DC coupled headphone amp on page 195.
SY, thanks for enquiring Morgan Jones directly. The man must not be happy you got his book, and most important, his e-mail
Erik
...what Morgan said...
Further to my comments about Morgan’s “dueling CCS” LTP, I looked up the data sheet (attached) for the MPSA92 high voltage PNP used as plate loads. The hfe is given by the data sheet as only 25 minimum, with no average or maximum stated. But Figure 1 shows a typical value of about 75. The emitter loads are 16K. So the collector impedance would be ((75+1) * 16K) + 1/hoe. Hoe is always a hard value to pin down, not only because it varies greatly unit-to-unit, but because it is non-linear. It is rarely specified. I’ll be generous and guess that 1/hoe = 100K. So the collector resistance could be just above 1 Meg, but it could just as easily be as low as a few hundred K too. In other words, as I said above, this is not a very constant “constant current source”, the only way it could be adjustable. While I’m on a roll (stop me…) take a look at Figure 2 in the data sheet. Variable capacitance – you knew I’d have to mention it.
Edit: Couldn't attach file. See link for Motorola data sheetMPSA92 data sheet
Further to my comments about Morgan’s “dueling CCS” LTP, I looked up the data sheet (attached) for the MPSA92 high voltage PNP used as plate loads. The hfe is given by the data sheet as only 25 minimum, with no average or maximum stated. But Figure 1 shows a typical value of about 75. The emitter loads are 16K. So the collector impedance would be ((75+1) * 16K) + 1/hoe. Hoe is always a hard value to pin down, not only because it varies greatly unit-to-unit, but because it is non-linear. It is rarely specified. I’ll be generous and guess that 1/hoe = 100K. So the collector resistance could be just above 1 Meg, but it could just as easily be as low as a few hundred K too. In other words, as I said above, this is not a very constant “constant current source”, the only way it could be adjustable. While I’m on a roll (stop me…) take a look at Figure 2 in the data sheet. Variable capacitance – you knew I’d have to mention it.
Edit: Couldn't attach file. See link for Motorola data sheetMPSA92 data sheet
In my phono preamp, I have two stages that are directly coupled. To prevent grid flashover on startup, the usual practice is to put a diode from grid to cathode of the second stage, clamping the voltage to 0.7. I sacrificed some protection and used an NE2-series lamp there (clamps at about 65-70V), which I am sincerely christening "The Brian Beck Clamp Lamp" in honor of the paranoia you've induced in me about 3 or 4 picofarads of modulation in circuits with impedances in the k-ohm range.
With your LED farm, and now your neons flickering at warm-up, you must have quite a light show in your listening room. Actually, the bootstrapping of the cathode makes a reverse-biased diode connected between grid and cathode ALMOST acceptable to the OCD-types like me, especially with higher-mu tubes.
Always happy to induce paranoia, though…
Brian “the Stephen King of DIYAudio” Beck.
Always happy to induce paranoia, though…
Brian “the Stephen King of DIYAudio” Beck.
Brian Beck said:
In my mic preamp it is 2.4 MOhm. Before that I used 10M resistor and additional source follower, but found that it is not necessary, without it a voltage on plates is stable enough. However, it depends on beta and temperature, but not so much, without servo anyway plate voltage is much less stable.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.