Alright, I spent the last two days optimizing the preamplifier. I initially tried the math approach, ie. trying to calculate exactly all the required component values etc. This however proved to be rather difficult; distortion and linearity did not seem to be consistent with my calculus. Hence I simply measured these parameters with RMAA and continued doing so with every component change I made and gradually improved the preamplifier.
The above schematic is the result of this painstaking process. As you can see the current for the first and second stage has increased considerably from the original Velleman schematic. The current for the voltage amplifier is now 6/2200 = 2.7mA and 152.5/2700 = 5.6mA for the cathode follower. Below are the results for the frequency reponse, noise level and the THD for both channels driven at 6Vpp output into a 22K load.
The frequency response needs some explanation, what you're looking at is the curve of my Audigy 2 ZS soundcard, the actual preamplifier has far greater bandwidth, but what's important here is to show that it does not affect the frequency response in any way, it is linear up to about 80KHz. These scores are acceptable for a tube design (as far as I can see). I'm sure they can be improved upon with other tubes and a different schematic, but this is as good as I got them without having to completely hack up the Velleman PCB.
Best regards,
Sander Sassen
http://www.hardwareanalysis.com
The above schematic is the result of this painstaking process. As you can see the current for the first and second stage has increased considerably from the original Velleman schematic. The current for the voltage amplifier is now 6/2200 = 2.7mA and 152.5/2700 = 5.6mA for the cathode follower. Below are the results for the frequency reponse, noise level and the THD for both channels driven at 6Vpp output into a 22K load.
The frequency response needs some explanation, what you're looking at is the curve of my Audigy 2 ZS soundcard, the actual preamplifier has far greater bandwidth, but what's important here is to show that it does not affect the frequency response in any way, it is linear up to about 80KHz. These scores are acceptable for a tube design (as far as I can see). I'm sure they can be improved upon with other tubes and a different schematic, but this is as good as I got them without having to completely hack up the Velleman PCB.
Best regards,
Sander Sassen
http://www.hardwareanalysis.com
And of course here's some pretty images to look at, as after reading all those posts you must've wondered what this preamplifier ended up looking like. You'll notice a few of the more obvious modifications, such as the upgrade from Wima MKS4 coupling caps to Rifa PHE450, which is a double metallized polypropyleen capacitor.
The keen eyed observer will also notice the switch from basic carbon to 1% metal film resistors and several other small modifications. Stock items supplied with the kit are the well known blue Panasonic volume control pot and the gold plated cinch connectors and the stainless steel backplate.
Best regards,
Sander Sassen
http://www.hardwareanalysis.com
The keen eyed observer will also notice the switch from basic carbon to 1% metal film resistors and several other small modifications. Stock items supplied with the kit are the well known blue Panasonic volume control pot and the gold plated cinch connectors and the stainless steel backplate.
Best regards,
Sander Sassen
http://www.hardwareanalysis.com
I notice that these are the JJ tubes. The plates look like they're rescued from ECC88s. The JJ ECC81s I've tested have remarkably lower distortion than the competition- I do wonder about the ECC82.
Have you run a loopback test on just the soundcard? I'm curious about how much of those higher harmonics are limitations of the measuring system (I also use an Audigy 2ZS).
Have you run a loopback test on just the soundcard? I'm curious about how much of those higher harmonics are limitations of the measuring system (I also use an Audigy 2ZS).
Hi Sander,
Your final values look quite reasonable, and should result in pretty good performance.
The JJ ECC82 seems to be the best of the current crop in my experience, but if you can get your hands on some dutch or german amperex they are even better.
The noise and distortion levels are quite good and should result in subjectively pleasing results..
How does it sound compared to the stock unit?
Your final values look quite reasonable, and should result in pretty good performance.
The JJ ECC82 seems to be the best of the current crop in my experience, but if you can get your hands on some dutch or german amperex they are even better.
The noise and distortion levels are quite good and should result in subjectively pleasing results..
How does it sound compared to the stock unit?
Hi Sander,
Maybe lifting the heater supply by 60V or something would be a good idea. Two more resistors, less stress on the heater/cathode isolation, probably even a bit less noise (stressed h-c isolation tends to add noise).
Tom
Maybe lifting the heater supply by 60V or something would be a good idea. Two more resistors, less stress on the heater/cathode isolation, probably even a bit less noise (stressed h-c isolation tends to add noise).
Tom
Hi Tom,
How would I go about accomplishing that?
Best regards,
Sander Sassen
http://www.hardwareanalysis.com
How would I go about accomplishing that?
Best regards,
Sander Sassen
http://www.hardwareanalysis.com
Hi Sander,
The distortion plot looks quite reasonable for such a tube set up. But it appears to me that –55 dB for the 2nd harmonic is still quite high for a pre amp at nominal output. Probably this is the best you can get with a ECC82?
Cheers 😉
The distortion plot looks quite reasonable for such a tube set up. But it appears to me that –55 dB for the 2nd harmonic is still quite high for a pre amp at nominal output. Probably this is the best you can get with a ECC82?
Cheers 😉
I think that's exactly right, it's the tube. At the same output level, a 6SN7 with no degeneration helping out shows 15dB better second, third just barely visible if at all, and everything higher order buried.
I guess it must be, I've tried various options (but for a CSS) to try and lower it, none of them successful. Even Morgan Jones' book explains that for a given tube you won't see much of an improvement unless you use a CSS. That obviously doesn't work for this particular design as the gain will equal mu, which is about 20 in this case.
Best regards,
Sander Sassen
http://www.hardwareanalysis.com
Best regards,
Sander Sassen
http://www.hardwareanalysis.com
Hi EC8010,
Just picked up 2x GE 12B4A’s. NOS from an electronic dump store. That was a nice suggestion, thanks. The mechanics inside the tubes look rather crude 😉 But it glows pretty nice orange hot with the cathode visible between the anode parts. Will try a Mu follower with mosfets with it.
Do you have any comments what about to take care off with this tube? And if it exits, can you point me to Pspce models for initial set up?
Regards
Just picked up 2x GE 12B4A’s. NOS from an electronic dump store. That was a nice suggestion, thanks. The mechanics inside the tubes look rather crude 😉 But it glows pretty nice orange hot with the cathode visible between the anode parts. Will try a Mu follower with mosfets with it.
Do you have any comments what about to take care off with this tube? And if it exits, can you point me to Pspce models for initial set up?
Regards
Sander,
You can post the PCB designer?
Change Wima to sonicap or audyn .
Change panasonic to nichicon mouse see:
http://db.audioasylum.com/cgi/m.mpl?forum=tweaks&n=92358&highlight=audyn&r=&session=
And congratulations!!! good job!!!
regards,
João
You can post the PCB designer?
Change Wima to sonicap or audyn .
Change panasonic to nichicon mouse see:
http://db.audioasylum.com/cgi/m.mpl?forum=tweaks&n=92358&highlight=audyn&r=&session=
And congratulations!!! good job!!!
regards,
João
Good first effort, will work. You don't need the diode at all, no historically similar design would have had this. The diode will always be reverse biased except if the first section fails in which case you'll be replacing the tube anyway or if you make a wiring error, or have a bad socket contact, short term operation with such a fault would be unlikely to destroy the tube. 😀 The diode depending on type may add a significant amount of fairly non-linear capacitance at the grid, although I assume it swamped by the comparatively low source impedance of the driving stage. Just not needed..
Gain can be reduced by removing the bypass cap. Lower gain might be achieved with SRPP topology.
Recommend somewhat higher currents for better linearity with the ECC82, a minimum of 5mA seems warranted, and even more is permissable in the follower section as long as you don't exceed the allowable section dissipation and overall dissipation limits for the tube.
I used to build a commercial pre-amplifier using ECC82 in an SRPP configuration with 820 ohm cathode resistors, 1K grid stopper resistors, +280V supply and a 2uF output capacitor. This tube is well suited to this topology. THD at normal signal levels (2Vrms) with a good tube was <0.05% .
Your output capacitor needs to be sized appropriately for the load impedance you are driving. I would recommend something on the order of 1uF minimum unless your amp has a high input impedance of >220K.
Gain can be reduced by removing the bypass cap. Lower gain might be achieved with SRPP topology.
Recommend somewhat higher currents for better linearity with the ECC82, a minimum of 5mA seems warranted, and even more is permissable in the follower section as long as you don't exceed the allowable section dissipation and overall dissipation limits for the tube.
I used to build a commercial pre-amplifier using ECC82 in an SRPP configuration with 820 ohm cathode resistors, 1K grid stopper resistors, +280V supply and a 2uF output capacitor. This tube is well suited to this topology. THD at normal signal levels (2Vrms) with a good tube was <0.05% .
Your output capacitor needs to be sized appropriately for the load impedance you are driving. I would recommend something on the order of 1uF minimum unless your amp has a high input impedance of >220K.
Kevin
Thanks for you very thorough remarks.
In fact, I am planning also to replace the decoupling capacitor by a 0.47 or 0.68 mF, size is an issue for this replacement.
Regards,
Felix
Thanks for you very thorough remarks.
In fact, I am planning also to replace the decoupling capacitor by a 0.47 or 0.68 mF, size is an issue for this replacement.
Regards,
Felix
SSassen said:
A small addition on the schema: you should add a resistor in parallel to the slider to earth. Otherwise, as the potentiometer gets older it will start getting poor contacts: in this case this will mean no conduction from grid to ground. This means that the gridcurrent forces the voltage up. You will heare this as cracking. A value of 330k is good in this case.
Secondly, I use a 10-15 pF between grid and anode of the input. With a 1k5 gridstopper. This results in a 'reduced' bandwidth of some 150 kHz with the Ecc82 mu of 10-20. (I use a 10 k potmeter) More than good enough. (otherwise it is >350 kHz @ -3 dB in my setting; without a cathode decoupling capacitor).
ps, sunsvc, the gridstopper should be seen as a low pass with the tiube capacitance * the mu, if I'm correct. So 1,6 pF * 20 = 32 pF. The source resistance is high in the case of a high valued potmeter.
regards
albert
kevin, I probably wouldn't skip the diode. It can prevent turn-on arcover. It was a standard feature in the DC stages of Tektronix scopes of the tube era.
Hi Sy,
I've never encountered a problem with arc-over in comparable circuits, but I suppose there is always a first time. Hasn't been a problem in any of the dc coupled amplifier circuits I have designed so far. (Caveat quantity limited.)
One big difference between the old tek scopes and this scenario was the presence of the dc delay relay which gave the 6080 pass tubes in the supply time to warm up before applying B+, I assume the dvdt at the moment of relay closing might be high enough to cause arc over in tubes with close element spacings like the 6DJ8 found in many stages of these scopes.
I generally use tube rectification and/or tube based series pass regulators so dvdt is on the order of tens of volts per second typically and under these conditions I would not expect arcing to be an issue.
No vintage commercial audio product I have encountered used diodes for protection (not commonly available until the 1960's) and I haven't yet encountered an ECC82 that failed this way. (I bet early silicon diodes cost more than the tubes they protected. LOL)
I am assume the diodes are in those Tek scopes to prevent long term reliability issues, and even possibly a quick and merciless death for the tubes they were intended to protect.
I figure it can't really hurt to leave it in provided that there is not significant leakage through it and if the capacitance changes are small as a consequence of very small variation in voltage impressed across it.
I avoid deliberately non-linear circuit elements, particularly ss ones whenever possible. Likely though that the circuit effects of that diode being there are not even measurable, so I am probably making much ado about nothing..
😀
I've never encountered a problem with arc-over in comparable circuits, but I suppose there is always a first time. Hasn't been a problem in any of the dc coupled amplifier circuits I have designed so far. (Caveat quantity limited.)
One big difference between the old tek scopes and this scenario was the presence of the dc delay relay which gave the 6080 pass tubes in the supply time to warm up before applying B+, I assume the dvdt at the moment of relay closing might be high enough to cause arc over in tubes with close element spacings like the 6DJ8 found in many stages of these scopes.
I generally use tube rectification and/or tube based series pass regulators so dvdt is on the order of tens of volts per second typically and under these conditions I would not expect arcing to be an issue.
No vintage commercial audio product I have encountered used diodes for protection (not commonly available until the 1960's) and I haven't yet encountered an ECC82 that failed this way. (I bet early silicon diodes cost more than the tubes they protected. LOL)
I am assume the diodes are in those Tek scopes to prevent long term reliability issues, and even possibly a quick and merciless death for the tubes they were intended to protect.
I figure it can't really hurt to leave it in provided that there is not significant leakage through it and if the capacitance changes are small as a consequence of very small variation in voltage impressed across it.
I avoid deliberately non-linear circuit elements, particularly ss ones whenever possible. Likely though that the circuit effects of that diode being there are not even measurable, so I am probably making much ado about nothing..
😀
I suspect that the reverse-biased diode contributes nothing bad. But if one does worry about such things, a neat trick from Marantz (from the mighty 9) was the use of an NE2 in the same position. Its firing voltage is about 80V, which is still low enough to prevent most arc-overs. And caacitance is super-low and quite linear.
- Status
- Not open for further replies.