I feel I have to apologise for my post yesterday, reading it now it seems a bit aggressive, I did not intend for it to come out that way.
I see that I was mistaken in what I wrote previously - the output voltage is going to be the product of the signal current and the load impedance (the latter term will vary in class B due to the feedback arrangement, but the former is constant on both sides of the differential pair due to the CCS), but it isn't going to be a concern if a different signal is given to a valve already in cutoff - its gain is zero and the signal is ignored.
I may ask a question (and I know this is getting slightly off topic):
The output impedance of the differential pair alters depending on the load: it is only equal to Ra||Rl if the outputs are balanced, so the output impedance of the differential pair will alter slightly as the output stage leaves its class A region of operation (in the class AB bias situation). The feedback ratio is dependent in part on the output Z of the differential pair, but Ra with pentodes is large - so, basically, would I be right in saying that this theoretical problem need not be a worry?
I had been thinking about losing the µ-followers and finally building it, but when sizing components, got sidetracked by what I thought were bad balance problems, and this was a convenient place to vent...
Eli Duttman said:
I see that I was mistaken in what I wrote previously - the output voltage is going to be the product of the signal current and the load impedance (the latter term will vary in class B due to the feedback arrangement, but the former is constant on both sides of the differential pair due to the CCS), but it isn't going to be a concern if a different signal is given to a valve already in cutoff - its gain is zero and the signal is ignored.
I may ask a question (and I know this is getting slightly off topic):
The output impedance of the differential pair alters depending on the load: it is only equal to Ra||Rl if the outputs are balanced, so the output impedance of the differential pair will alter slightly as the output stage leaves its class A region of operation (in the class AB bias situation). The feedback ratio is dependent in part on the output Z of the differential pair, but Ra with pentodes is large - so, basically, would I be right in saying that this theoretical problem need not be a worry?
Sorry for hijacking your thread. I mused over making a circuit similar in principle to the one that Bandersnatch posted in this thread three or four years ago using 8116/YL1071 double tetrodes as outputs. The schematics are still here (It's some crazy thing involving differential pentode based µ-followers direct couped to the output, which I wouldn't build now, so be warned - it's quite scary).
I had been thinking about losing the µ-followers and finally building it, but when sizing components, got sidetracked by what I thought were bad balance problems, and this was a convenient place to vent...
clip/quote----I may ask a question (and I know this is getting slightly off topic):
The output impedance of the differential pair alters depending on the load: it is only equal to Ra||Rl if the outputs are balanced, so the output impedance of the differential pair will alter slightly as the output stage leaves its class A region of operation (in the class AB bias situation). The feedback ratio is dependent in part on the output Z of the differential pair, but Ra with pentodes is large - so, basically, would I be right in saying that this theoretical problem need not be a worry?
hey-Hey!!!,
I have not gone and measured the change in input Z of a tube cut off v. conducting. So, assume it changes a bit. Make its grid resistor as small as is reasonable, make the LTP plate loads small also and a small difference won't make itself visible over top of those effects.
cheers,
Douglas
The output impedance of the differential pair alters depending on the load: it is only equal to Ra||Rl if the outputs are balanced, so the output impedance of the differential pair will alter slightly as the output stage leaves its class A region of operation (in the class AB bias situation). The feedback ratio is dependent in part on the output Z of the differential pair, but Ra with pentodes is large - so, basically, would I be right in saying that this theoretical problem need not be a worry?
hey-Hey!!!,
I have not gone and measured the change in input Z of a tube cut off v. conducting. So, assume it changes a bit. Make its grid resistor as small as is reasonable, make the LTP plate loads small also and a small difference won't make itself visible over top of those effects.
cheers,
Douglas
30W EL84 PP good article
http://lundahl.se/pdfs/claus_byrith/amplifier_30wpp.pdf
http://lundahl.se/pdfs/claus_byrith/appendix_cb.pdf
/sreten.
http://lundahl.se/pdfs/claus_byrith/amplifier_30wpp.pdf
http://lundahl.se/pdfs/claus_byrith/appendix_cb.pdf
/sreten.
Shoog said:
hey-Hey!!!,
I have tried it a few times( just to be sure ), and there is too much FB. There is no way to swing the required voltage at the grids. Maybe with a McIntosh-style Bifilar and run at double B+ it might but that just adds needless complexity and trouble I think. Besides, you don't need that much FB. Tapping off the 20% taps is quite sufficient...
that's why I get Heyboer to wind me OPT with more than one pair of taps; g2 usually likes more percentage than is useful for the E-Linear connection.cheers,
Douglas
One question that comes to mind with Bandersnatch's e-linear schematic from the beginning of this thread:
Is this only one channel of the amp? And if so, were you building monoblocks? Because what I was shooting for was a stereo amp, possibly integrated with the preamp built in.
So if I am reading it correctly, the amplifier section would have 2) 6AU6 and 2) EL34 per channel.
Just trying to understand.
Thanks!
Is this only one channel of the amp? And if so, were you building monoblocks? Because what I was shooting for was a stereo amp, possibly integrated with the preamp built in.
So if I am reading it correctly, the amplifier section would have 2) 6AU6 and 2) EL34 per channel.
Just trying to understand.
Thanks!
SigloOne said:
hey-Hey!!!,
That is indeed one channel shown, and you have the tube count right. This will fit into a small Dynaco St.70 chassis even. I have two sets of monoblocks in almost this exact circuit( I used the 6AC7 instead of the 6AU6 ). One pair built from Dynaco Mk.III and the other from Heathkit W6m.
For an integrated, I would put a single triode ahead of it, perhaps a 12B4 as not much additional gain would be needed. Resistive loaded and R-C coupled( aka cap coupled ) would do quite well.
cheers,
Douglas
Any rough idea of cost for the parts to build this? As just an amp? Just the pre-amp? or as an integrated amp?
I need to get an idea of what I may need to budget. I am not flush with cash, and I have to take that into consideration.
Also do you have a circuit for the preamp? How many transformers are we talking if I go with the integrated design (which is what I would prefer), can it be done with 1 Mains transformer and two output transformers, or are more needed?
I need to get an idea of what I may need to budget. I am not flush with cash, and I have to take that into consideration.
Also do you have a circuit for the preamp? How many transformers are we talking if I go with the integrated design (which is what I would prefer), can it be done with 1 Mains transformer and two output transformers, or are more needed?
SigloOne said:
The list of magnetics is: a power trafo, a filter choke, and a pair of O/P trafos. An AnTek AN-4T360 inexpensively takes care of all power needs, except the bias supply. Greinacher ("full wave") voltage doubling a "Rat Shack" catalog # 273-1366 deals with that requirement, in a cost effective manner. Buy a pair of the $62.27 Edcor CXPP60-8-4.2K O/P trafos. Hammond's model 159T satisfies the choke need. Perhaps Edcor can competetively wind a similar part.
I'll let Doug provide exact details, but the 6AU6 differential pair should be capable of providing all the requisite gain, if a CDP is the source.
Converting Doug's schematic into an "integrated" amp primarily means replacing the I/P 6AU6 grid leak resistor with an Alpha Taiwan 100 KOhm log. taper pot. from Mouser. Individual pots. for each channel are the most cost effective way of obtaining control over both listening level and channel to channel balance.
BTW, adding a LITTLE GNFB to Doug's design, for budget O/P trafo linearization purposes, is not especially complicated.
Forgot to say, my current amp which uses 807's in PP pentode mode, is predicted to deliver 25W in pure class A so it could be a candidate. However its not a build for the faint hearted. Never tested its true limits because my speakers only need a few watts before they start breaking up.
Shoog
Shoog
hey-Hey!!!,
If you're going to delete the linestage or additional stage of gain, go with the 6AC7 front end. It will take up more room( octal v. 7 pin mini sockets ). You'll realize a bit more gain and in this case it is worth doing.
The bias can be done quite nicely with a 450 mA/25V $6 RadioShack tx. Doubler it, take the midpoint voltage for the CCS and full for the bias. They're small too.
I would go and get the best OPT you can afford. The 5k Hammond 1650R is fantastic, and Heyboer TX in Michigan can do some excellent ones too.
cheers,
Douglas
If you're going to delete the linestage or additional stage of gain, go with the 6AC7 front end. It will take up more room( octal v. 7 pin mini sockets ). You'll realize a bit more gain and in this case it is worth doing.
The bias can be done quite nicely with a 450 mA/25V $6 RadioShack tx. Doubler it, take the midpoint voltage for the CCS and full for the bias. They're small too.
I would go and get the best OPT you can afford. The 5k Hammond 1650R is fantastic, and Heyboer TX in Michigan can do some excellent ones too.
cheers,
Douglas
If I stay with the integrated design, can stepped attenuators be used in place of the taper pots? or do you think the taper pots are the better choice?
Also, what do you recommend for a source selector switch? I want something that will be reliable and not "pop" when switching sources. I need inputs for CD, I-Pod, Home Theater, Tape, and phono. I do have a modern phono pre-amp, but it's not tube based. Would like to take that out of the loop if I can.
Are tape outs possible, or does that get even more complicated.
And it's not that I want to cheap out on the OPT's, I do want quality, just don't want to get hosed.
Thanks!
Also, what do you recommend for a source selector switch? I want something that will be reliable and not "pop" when switching sources. I need inputs for CD, I-Pod, Home Theater, Tape, and phono. I do have a modern phono pre-amp, but it's not tube based. Would like to take that out of the loop if I can.
Are tape outs possible, or does that get even more complicated.
And it's not that I want to cheap out on the OPT's, I do want quality, just don't want to get hosed.
Thanks!
SigloOne said:
I think that you will find that a consensus is developing to the effect that Edcor "iron" is a better value than Hammond. If you want to move up a notch, buy the now on sale at $89.95 DynaClone A470 from Triode Electronics. Dyna style O/P trafos have been price/performance champions for a long time.
A non-shorting (break before make) switch is appropriate in the selector role. Mouser part # 105-14572 will do the job well, without bankrupting you.
BTW, you should always turn the volume controls down, when changing signal sources.Hot molded Carbon controls by PEC cost less than stepped attenuators, while sounding better.
IMO, PEC pots. are the logical step up from Alpha Taiwan. Buy PEC controls from DigiKey.It's perfectly OK to use the SS phono preamp until you build a tubed "box".
Adding a set of recording O/Ps is easy, but doing it "right" involves some complications. Many recording devices present a nasty low impedance load, particularly in the powered off state. Dealing with that reality suggests that the recording O/Ps be buffered. "P" channel JFET source followers powered by the negative rail will get the job done. The SS buffers get connected to the O/Ps of the source selector switch in parallel with the I/Ps of the LTPs.
How much of a difference are the DynaClone transformers going to make over the Edor's. They price out at Dyna's $300 and Edcors $200. Pretty big price difference. Is there going to be a noticeable performance increase?
BTW: I think I will forget tape outs, since I hardly use that anymore.
Another thing: what to do about chassis. I have looked for more "dressier" chassis than the ones Hammond makes, but have found little. It seems when i search for chassis, most everyone carries the Hammonds. I do have a friend at a machine shop who can help me custom make a chassis out of aluminum or copper. but it could be a lot of work. I saw some nice ones at thlaudio.com, but with shipping and all, two of them would have been over $600, yikes.
Any suggestions.
BTW: I think I will forget tape outs, since I hardly use that anymore.
Another thing: what to do about chassis. I have looked for more "dressier" chassis than the ones Hammond makes, but have found little. It seems when i search for chassis, most everyone carries the Hammonds. I do have a friend at a machine shop who can help me custom make a chassis out of aluminum or copper. but it could be a lot of work. I saw some nice ones at thlaudio.com, but with shipping and all, two of them would have been over $600, yikes.
Any suggestions.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.