| oshifis |
Currently I am using capacitor shunted cathode resistors for setting the DC operating point (807 triode connected, 425V B+, 560 ohms parallel 2 x 10uF / 160V) that sets about 50...60mA.
What happens if I connect a constant current source, like an LM317 with a 25 ohms current sensing resistor in place of the 560 ohms cathode resistor? How do I calculate the necessary shunt capacitor? Does it make sense at all? |
|
|
| Shoog |
It makes most sense in a PP amp where it fixes the current through the output transformer, eliminating saturation currents from the core.
In SE amps, as you are describing, with fully bypassed cathodes - I would say that it probably makes absolutely no difference since all of the AC signal is going through the cap in either case.
Still my RH807 SE amp is setup to use LM317 CCS in the cathode and it certainly does no harm.
Shoog |
|
|
| oshifis |
| I intend to use it in a push-pull amplifier, but I've never seen any circuit diagram using a CCS in the output tubes' cathode. My problem is with the shunting capacitor. The CSS is virtually an infinite resistor, so the determining element is the output impedance of the cathode, right? How do I get it from the curves? |
|
|
| d2134 |
| Use the same shunting capacitor as for the 560 ohm resistor. |
|
|
| cerrem |
It does not make much sense to me, if your using a common tied cathode from both output tubes...
Saturating currents in the transformer occur from IMBALANCE in the output tube pair in P-P amps.... Since the DC cancels when it is balanced...
Unless you are using a "seperate" CCS for each output tube independently.... Then you would bypass each with a cap based on the resistance looking into each single cathode, which is basically 1/gm ..... Once you know the R, then you set the time constant for a -3dB POLE at no higher than 6 Hz....
In common cathode P-P outputs, you would not have an AC signal at the cathode due to cancelation, but in real life the tubes do not track with perfect balance, which makes for small degenerative feedback, which may be good...or use a bypass cap to bypass what it not canceling...
Chris |
|
|
| Johan Potgieter |
| If output operation is studied it will show that there is mostly an increase in current near max. output and for lowest distortion. A CCS will thus limit max. output power to a little lower than obtainable with cathode bias. |
|
|
| Shoog |
If you use a single CCS for an output pair you need to include a current balancing pot to null current imbalances and minimise static core saturation.
The better way is to have independent CCS for each cathode and bypass them seperately. This can be turned into a differential pair by connecting the two caps cathodes together and referencing the node to a 1meg resistor, which then goes to earth. This unfortunately puts your caps in series and so effectively halves their value - so a cap of twice the size is needed. Bypass the caps with film caps and bypass each valve cathode to the other with a film cap. This works very well and needs no rebalancing as the valves age.
Shoog |
|
|
| oshifis |
| Thanks for the ideas to all. I tried independent CCS in each cathode (an LM317 with 25 ohms current setting resistor, 50 mA idle current). The bypass capacitor is 2 x 10 uF polypropylene. It works, but the available output is lower than with autobias. I think this is because there should be a minimum 3 V voltage drop on the CCS, which reduces the g1 voltage swing. The voltage drop is about 36 V on the CCS. So the maximum voltage swing is 33 V peak or 23 V RMS. I could drive the output tubes in autobias with 26 V RMS indeed. |
|
|
| 7N7 |
A constant current sink in the output stage of a p-p amplifier, apart from any direct advantages is a nice idea for DC-coupled amplifiers.
I once built a DC-coupled amplifier using a 6528A as the output pair I sat this on top of a 12E1 beam tetrode as a sink. I recall that the amplifier sounded very nice.
7N7 |
|
|
| Shoog |
The full bias voltage should be availble to swing. Since the cathodes are fully bypassed the AC signal has absolutely no dealings with the LM317. All AC signal is going through the bypass cap.
The choice of 10uf caps is totally inadequate. You will get a very bright presentation, and the area which loudness is most noticable will be attenuated.
In my 6080 amp I use 2000uf on each cap. This gives an effective bypass value of 1000uf because I use the differential configuration described in my last post. The 6080 is extremely low output impedance so a big cap is needed. I would say that you should be starting to look at values of at least 470uf and probably bigger than 1000uf. You can keep your 10uf caps as bypasses to the big electro's. Don't try to comment on the final result until those bypasses have been massively bumped up.
Shoog |
|
|
| oshifis |
| quote: | Originally posted by Shoog
I would say that you should be starting to look at values of at least 470uf and probably bigger than 1000uf. |
Shoog,
Thanks for your comments. I will try the big cap. What type do you recommend? I am not a big fond of electrolytics.
I also tried the autobias 560 ohms resistors in each cathode without bypassing capacitor. Although the gain was less, it was not so much less. I guess the local negative feedback gives the most linear operation.
Laszlo |
|
|
| Shoog |
You are obviously going to have to use electro's. Try it with whatever ones you have lying around before spending any money on fancy caps - you may not like the result.
I think if you go for the differential approach I described, you can expect the advantages to go a long way to reducing your concerns about the electros. For me the improvement over a CCS with basic electro bypassing was so huge I haven't looked back.
Shoog |
|
|
| Johan Potgieter |
Just another thought, in support of Shoog.
Some will frown at my insensitivity to basic science, but I also think the concern about electrolytic capacitors is overrated in audio. Also the attributed superiority of exotic (and expensive!) electrolytics. One is not saying these caps are perfect, but modern ones end up doing the job without mentionable added effects. (There is some sense in splitting up very large Cs into smaller units.) |
|
|
| Shoog |
I wouldn't go so far as to say that caps don't sound different. My latest amp is sufficiently simple that changing the one parafeed cap in it can have a significant impact on the sound characture. I ended up using a paper in oil cap with polystyrene bypass.
I would say that cathode bypass caps don't seem to make as much difference as coupling caps. It is cheap and easy to place film bypasses over big electros to "tune" there sound. I do think that if you go for caps designed for SMPS you can expect them to be the best sounding in cathode bypass duty.
I often see people been so afraid of electros in cathode bypass duty that they underspecify them, and so introduce bass roll off - this is certainly a worse result than slamming in a big electro.
Using the differential approach I describe (and which Brian Beck suggested to me) halves the ESR and other parasitics of the big caps.
Overall I think that the choice of circuit will have a massively bigger effect on the end result than your choice of components. This is why I say that the differential stage brings the nasties of the cap to the sound, but improves the sound to such an extent that those nasties are well and truly swamped.
It amazes me that people spend so much energy thinking over components and then slam them into a bog stand off the shelf amp designs - which they poorly understand. There are so many high performing "exotic" circuits out there which wipe the floor with the more traditional SE and PP designs that it defys logic why people still build the old designs. Take a top flight circuit and build it out of salvaged cheap components - and its still top flight !!! Expensive components are often a crutch for a poor design.
That about sums up my philosophy - RANT OVER.
Shoog |
|
|
| 7N7 |
| quote: | Originally posted by Johan Potgieter
If output operation is studied it will show that there is mostly an increase in current near max. output and for lowest distortion. A CCS will thus limit max. output power to a little lower than obtainable with cathode bias. |
Correct me if I am wrong, but this surely only applies when an output stage is moving from Class A to AB?
So my 6528 amplifier I mentioned above, must have been guaranteed to be only Class A, but it did make 10W I recall which is quite surprising on reflection.
I must see if I still have the circuit diagram!
7N7 |
|
|
| oshifis |
| quote: | Originally posted by oshifis
I also tried the autobias 560 ohms resistors in each cathode without bypassing capacitor. Although the gain was less, it was not so much less. I guess the local negative feedback gives the most linear operation. |
I measured a gain drop of 6 dB without shunt capacitor. The grid AC voltage is 26 V, the OT is 27.5:1 and the secondary is 8 V with capacitor, 3.9 V without capacitor. I did not see any significant decrease in THD at the same output power. |
|
|
| arnoldc |
I used "expensive" parts when I get them at a bargain. For example, the Tamura OPTs on my 45 came with a free Tamura choke. The MagneQuest DS-025 was sold to me at a "special" price. I stop there.
When I prototype any project, they use the cheapest, commonly available parts. The most exotic parts I've used are Auricaps, Jensen and Mundorf for coupling. Black Gate for bypass (my friend is the "pusher" here) and Audio Professor for PSU.
I second Shoog on the circuit. While I consider myself a newbie in brewing circuits, I have this forum for support. So far, I liked what I came up with, and there's been a couple already! |
|
|
| Johan Potgieter |
| quote: | Originally posted by 7N7
Correct me if I am wrong, but this surely only applies when an output stage is moving from Class A to AB?
|
7N7,
Theoretically so, but there is a small increase in total current at maximum output in pure class A. This is explained in Langford-Smith; pardon me for being lazy now (it's cold out here!) - I would rather look it up and come back to you than waffle on here now. Data books also show this.
Shoog,
I would say that capacitors need have no influence on sound. (That is, obviously, not talking about stupid choices like small ceramics for filter circuits, etc.) This was amply illustrated in an excellent series of articles by Cyril Bateman in Electronics World, August 2002 - ? (Everybody should read these.) Further illucidating, that there is no reason for exotic capacitors - polyester/polypropilene types are more than adequite for coupling etc. in audio.
Repeating, I also get itchy when the virtues of overpriced electrolytic capacitors are extolled. And the claim of "no electrolytics in the signal circuit" makes me reach for antacid! There cannot be an electrical principle much more basic than that current flows in a circle. Thus the power supply filter caps and any other decoupling caps are as much in the "signal circuit" as anything, and I have never seen a circuit without at least one of those.
There, let me calm down and simply agree with your post. |
|
|
| Shoog |
| quote: | | Theoretically so, but there is a small increase in total current at maximum output in pure class A. This is explained in Langford-Smith; pardon me for being lazy now (it's cold out here!) - I would rather look it up and come back to you than waffle on here now. Data books also show this. |
The basic assumption with a differential output stage is that it can only work in pure class "A". This is logical and what it is designed to do. If you need class AB1 or AB2 don't try to implement a differential.
Shoog |
|
|
| 7N7 |
| quote: | Originally posted by Johan Potgieter
7N7,
Theoretically so, but there is a small increase in total current at maximum output in pure class A. This is explained in Langford-Smith; pardon me for being lazy now (it's cold out here!) - I would rather look it up and come back to you than waffle on here now. Data books also show this.
|
Johan & Shoog,
Thanks for your responses.
At the moment I have been drawing up some ideas. I have long thought that since I like diff pairs anyway, then the output stage should conform to this ideal. Last night's effort involves three stages with each on a cascode CCS. The output stage (triode strapped YL1071s) will pass 180mA.
To save money, the idea is an isolation transformer to provide about 320V HT and a smaller toroid to give -80V, thus providing (including bias) about 340V for the output stage and a nice -ve rail for the other sinks.
The amplifier is not otherwise very interesting. I might try though to re-draw it to be all DC-coupled, but that will take a bit of time as I am a little "rusty"!
7N7 |
|
|
| oshifis |
I tried three versions with 8 V RMS output @ 8 ohms load:
1.) 560 ohms resistors in each cathode. k = 1.45%
2.) 560 ohms with 470 uF in each cathode. k = 0.26%
3.) 50 mA current generator (LM317) with 470 uF in each cathode. k = 1.22%
So the best option is what Shoog advised. I don't understand why the CCS gave higher distortion than a simple resistor. |
|
|
| Shoog |
| quote: | 1.) 560 ohms resistors in each cathode. k = 1.45%
2.) 560 ohms with 470 uF in each cathode. k = 0.26%
3.) 50 mA current generator (LM317) with 470 uF in each cathode. k = 1.22%
So the best option is what Shoog advised. I don't understand why the CCS gave higher distortion than a simple resistor. |
Seems like the capacitor is the dominate distortion generator.
Have you tried it with film bypasses. Its also a good idea to place a film cap directly between the two PP cathodes.
Since you seem to have a better measurement setup than me, I would be interested if you could lash up the differential setup and give us a distortion measurement.
Shoog |
|
|
| Shoog |
| EDIT: I just thought - the simple resistor is introducing degenerative feedback, and so linearising the valves output. The CCS has no degenerative feedback so it is bound to have slightly higher distortion. The cap probably has very little to do with it. |
|
|
| yagoolar |
| quote: | Originally posted by oshifis
I tried three versions with 8 V RMS output @ 8 ohms load:
1.) 560 ohms resistors in each cathode. k = 1.45%
2.) 560 ohms with 470 uF in each cathode. k = 0.26%
3.) 50 mA current generator (LM317) with 470 uF in each cathode. k = 1.22%
So the best option is what Shoog advised. I don't understand why the CCS gave higher distortion than a simple resistor. |
4) 50 mA CCS without a cap - k=? |
|
|
| oshifis |
| quote: | Originally posted by Shoog
Seems like the capacitor is the dominate distortion generator.
Have you tried it with film bypasses. Its also a good idea to place a film cap directly between the two PP cathodes.
Since you seem to have a better measurement setup than me, I would be interested if you could lash up the differential setup and give us a distortion measurement.
Shoog |
The distortion was lower with bypassing capacitor, perhaps because of the higher voltage gain that needed lower driving voltage. In unbypassed configuration the driving voltage was 53 V RMS, with bypassed it was 26 V (for the same 8 V output).
My measurements were done with 20 uF polystyrene capacitors (WIMA MKS-4) bypassing ROE 470 uF/63V electrolytic caps.
I will try the differential setup, but I need to modify the circuit a bit. Now I have a center tapped secondary winding, and the lower ends of the bias resistors (or CCS) are connected to each end of the secondary (to the speaker terminals) cross-coupled. This is approx a 1.5 dB negative feedback. Since it does not do much, I will try without it.
I did not try the CCS without bypassing. Will I get any gain? |
|
|
| yagoolar |
| Just curious about it. |
|
|
| Shoog |
| quote: | | I did not try the CCS without bypassing. Will I get any gain? |
No
Shoog |
|
|
| yagoolar |
| quote: | Originally posted by Shoog
No
Shoog |
What about distortion? |
|
|
| oshifis |
| No gain -> no distortion :-) |
|
|
| yagoolar |
| quote: | Originally posted by oshifis
No gain -> no distortion :-) |
That is quite brave thesis. :eek: |
|
|
| 7N7 |
| quote: | Originally posted by yagoolar
That is quite brave thesis. :eek: |
Yes and not true...!
7N7 |
|
|
| yagoolar |
| quote: | Originally posted by 7N7
Yes and not true...!
7N7 |
AFAIR gain is related to linearity ... :confused: |
|
|
| 7N7 |
| quote: | Originally posted by yagoolar
AFAIR gain is related to linearity ... :confused: |
No, I was agreeing with you - for example as far as I am aware, a resistor does not have "gain" as I understand it , but I know that it can introduce distortion!
7N7 |
|
|
| Circlotron |
| quote: | Originally posted by oshifis
No gain -> no distortion :-) |
A complementary symmetry emitter follower output stage has no [voltage] gain. There are lots of people that would be happy to know that these produce no distortion. :) |
|
|
| Circlotron |
| As a valve ages, a ccs would adjust the bias to ensure the cathode current remains constant as the cathode material gradually wears out. This would have to be a good thing. |
|
|
| yagoolar |
| quote: | Originally posted by yagoolar
AFAIR gain is related to linearity ... :confused: |
Gosh! I meant distortion, NOT gain! |
|
|
| Dave Cigna |
yagoolar, oshifis is using a separate CCS under each tube. If those CCS's were unbypassed then there could be no AC signal current through any of the tubes. The current would be constant DC no matter what signal was applied to the grids.
The voltage at the cathodes would respond to signal at the grids (cathode followers!) but there would be no AC signal current at the plates for the output transformer to respond to; there would be zero output at the speaker.
-- Dave |
|
|
| yagoolar |
| quote: | Originally posted by Dave Cigna
yagoolar, oshifis is using a separate CCS under each tube. If those CCS's were unbypassed then there could be no AC signal current through any of the tubes. The current would be constant DC no matter what signal was applied to the grids.
The voltage at the cathodes would respond to signal at the grids (cathode followers!) but there would be no AC signal current at the plates for the output transformer to respond to; there would be zero output at the speaker.
-- Dave |
Hm. If we take parafeed setup with a CCS in an anode there wouldn't be AC signal either.
The other case is when a LED is in cathode circuit. Not quite a CCS, bur still ...
Am I missing anything?
Pawel |
|
|
| Shoog |
| quote: | Hm. If we take parafeed setup with a CCS in an anode there wouldn't be AC signal either.
The other case is when a LED is in cathode circuit. Not quite a CCS, bur still ...
Am I missing anything?
Pawel |
Indeed you are missing something.
With the CCS in the cathode it stops AC developing over the cathode and so imposses a fixed AC bias. It acts like an overwhelming amount of degenerative feedback (imagine what would happen if you put a one meg resistor in the cathode of a tube - but allowed the DC bias current to flow as normal ).
With the CCS in the anode it acts as a very stiff load (infact a perfect load) which the plate pumps up and down on. Since the plate is not the controlling element, the plate is free to move. The valve is allowed to realise all of its mu as gain rather than having it absorbed by the load resistor.
An LED is a fixed voltage reference with an extremely low impedance - in fact it is the complete opposite of a CCS.
Shoog |
|
|
| oshifis |
| quote: | Originally posted by Shoog
The better way is to have independent CCS for each cathode and bypass them seperately. This can be turned into a differential pair by connecting the two caps cathodes together and referencing the node to a 1meg resistor, which then goes to earth. This unfortunately puts your caps in series and so effectively halves their value - so a cap of twice the size is needed. |
I have cross-coupled secondaries (center tapped to GND) to the bottom of the cathode resistor/bypass capacitor. This effectively forces the cathode AC voltages equal but opposite phase. The capacitor coupled cathodes make the cathode AC voltages equal in phase. I did not try the current generator/capacitor coupled cathodes yet. Now I am happy with the result. I use the original 560 ohm cathode resistors and 2200 uF bypass capacitors (shunted with 20 uF polystyrene). The lower end goes to the output transformer secondaries in opposite phase for both cathodes. |
|
|
| batinal |
Dear All,
I am planning to mod my small pentode PP amp to use CCS(10M45, U5 in the schematic) replacing 200R resistor. Is it going to work and what kind of improvements can I expect, electronically and accoustically? The cathode current present is 72 mA at 14.3 V.
Thank you for your help.
Kay |
|
|
| Shoog |
| quote: | I am planning to mod my small pentode PP amp to use CCS(10M45, U5 in the schematic) replacing 200R resistor. Is it going to work and what kind of improvements can I expect, electronically and accoustically? The cathode current present is 72 mA at 14.3 V.
Thank you for your help. |
Yes it should work. Whats R8 for ??
Understand that you are restricting the amp to pure class A operation.
I would recommend using two CCS, one for each cathode. This will make hum a thing of the past, and should improve the quality of the bass.
If you use just one CCS, people have reported improvements in definition and bass. You will want a bias balancing network to get the most benefit. ala the Simple EL84 amp on "diyparadise" (you should find a link to this in this thread).
Of course if you do decide to use two CCS, I recommend setting them up as a fully differential pair as described previously.
Shoog |
|
|
| batinal |
Hello Shoog,
Thanks for prompt reply with lots of help.
Sorry, I should've deleted R8.
Best regards,
Kay |
|
|
|
