Hi Guys,
I’ve been fooling around with this preamp for a while and would appreciate your experienced opinions before I start drilling holes in my chassis.
I started with the transformer coupled linestage schematic on the JE labs website, but quickly became hooked by the “Ultrapath” topology.
The main problem is that the B+ from the JE labs power supply is way too high and I am unsure of the best way to bring it down. I added an RC stage to the end of the supply, but wonder if it would be better to add more chokes, or try a VR regulator tube? (Note: I use a Mac, and am unable to do simulations in PSUD2)
Of course, I would appreciate any other suggestions you may have!
Thanks,
Matt
Here are some pictures of the linestage:
Voltage at the first 500uf capacitor is 291.5VDC, at the second 500uf 290.1VDC, at the 43uf Solens 276.3VDC and at the Ultrapath cap 271VDC. I end up with 251.5VDC on the plates once I subtract the bias voltage (Design centre max for the 76 is 250V).
I’ve been fooling around with this preamp for a while and would appreciate your experienced opinions before I start drilling holes in my chassis.
I started with the transformer coupled linestage schematic on the JE labs website, but quickly became hooked by the “Ultrapath” topology.
The main problem is that the B+ from the JE labs power supply is way too high and I am unsure of the best way to bring it down. I added an RC stage to the end of the supply, but wonder if it would be better to add more chokes, or try a VR regulator tube? (Note: I use a Mac, and am unable to do simulations in PSUD2)
Of course, I would appreciate any other suggestions you may have!
Thanks,
Matt
Here are some pictures of the linestage:
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
Voltage at the first 500uf capacitor is 291.5VDC, at the second 500uf 290.1VDC, at the 43uf Solens 276.3VDC and at the Ultrapath cap 271VDC. I end up with 251.5VDC on the plates once I subtract the bias voltage (Design centre max for the 76 is 250V).
Depending on operating point the Rp of a 76 varies between 9.5 - 12K typically. Depending on the quality of the opt (whether or not there is enough primarly L) the bass could start to roll off early in this particular application. I would use a 15K - 20K plate to line transformer with several hundred H primary Z, better still in this case might be parafeed with a ccs..
I use a 26 (8K Rp) with Ha-133 15K:600 plate to line transformers into 70K input impedance and I have good performance to both extremes. With these transformers I do get down to slightly below 30Hz and to something in excess of 25KHz (IIRC) on the top end with better than +/-1dB - way better as I recall.
I was suspicious of the whole transformer coupled gig for quite a while after I built this one, it was just in comparison to a number of other pre-amps I owned or had access to that this one finally won out.
Incidentally an excellent choice for dht based line stages is the 12A/112A, its Rp is low enough for good performance with a 10K transformer. The 6J5/6SN7 is another excellent choice..
I use a 26 (8K Rp) with Ha-133 15K:600 plate to line transformers into 70K input impedance and I have good performance to both extremes. With these transformers I do get down to slightly below 30Hz and to something in excess of 25KHz (IIRC) on the top end with better than +/-1dB - way better as I recall.
I was suspicious of the whole transformer coupled gig for quite a while after I built this one, it was just in comparison to a number of other pre-amps I owned or had access to that this one finally won out.
Incidentally an excellent choice for dht based line stages is the 12A/112A, its Rp is low enough for good performance with a 10K transformer. The 6J5/6SN7 is another excellent choice..
Like this ?
An externally hosted image should be here but it was not working when we last tested it.
How do I calculate the value of this resistor? (I'm just a hobbyist with one term of engineering under his belt
)I was under the impression that the 6SN7 and 76 were very simmilar - would it be better to use a 6SN7 with this transformer? I did most of my listening through a pair of Senheisser HD600's and while I felt the high end was slightly rolled off, the bass seemed fine.
Here's a copy of the data sheet for the transformer if it helps!
An externally hosted image should be here but it was not working when we last tested it.
Thanks,
Matt
Exactly.
It will either be correct or not. If not, change the polarity of the 6.3V winding. You can do some simple tests beforehand to be sure you got it right the first time.
The method you choose really depends on your end goal. I don't think you mentioned what voltage you were shootin' for?
It will either be correct or not. If not, change the polarity of the 6.3V winding. You can do some simple tests beforehand to be sure you got it right the first time.
The method you choose really depends on your end goal. I don't think you mentioned what voltage you were shootin' for?
hpn5350 said:How do I calculate the value of this resistor? (I'm just a hobbyist with one term of engineering under his belt
Download this baby: http://www.duncanamps.com/psud2/index.html
It's free, it works (very accurately if you measure your transformer primary and secondary resistance, and unloaded primary and secondary voltage). Best of all, you can plug in a bunch of different values and begin to get a good feel for the relative effect of various parameters.
Oh, and it's real easy to use. Just ask here if you have any questions.
Sheldon
Actually the 76 and 6J5/6SN7 family are not all that similar.
Depending on operating point the 76 has an Rp of 9.5K - 12K typical, a gm of 1150 - 1450 umhos, and a mu of 13.8.
The 6J5/6SN7 family depending on operating point will have an Rp of about 6.5K - 8K, a gm of 2600 - 3000 umhos, and a mu of 20.
I believe the early 6J5G is actually a lineal descendent of the 76 - all of which can trace back to the 27, the first idht introduced in the late 1920's.
Other obvious differences are better (perhaps unstated) cathode construction and materials, better gettering, and octal bases allowing the use of a universal socket design for lots of different tube types.
IMHO The 76 is not the best match in terms of Rp that you could make to the 10K James transformer. I think the 6J5G or 6SN7 would be a better choice. Other types with somewhat to considerably lower Rp would be even better.
5687, 5842/417A are other types I would consider for this application as long as Ip is kept to slightly below 20mA to avoid saturation issues in the transformer.
Note you will get better low frequency extension with tubes with an Rp below 10K when using the 10K tap - you will also probably get less distortion as well. Kind of the point with my harping on the Rp issue..
Edit: kept adding thoughts LOL
Depending on operating point the 76 has an Rp of 9.5K - 12K typical, a gm of 1150 - 1450 umhos, and a mu of 13.8.
The 6J5/6SN7 family depending on operating point will have an Rp of about 6.5K - 8K, a gm of 2600 - 3000 umhos, and a mu of 20.
I believe the early 6J5G is actually a lineal descendent of the 76 - all of which can trace back to the 27, the first idht introduced in the late 1920's.
Other obvious differences are better (perhaps unstated) cathode construction and materials, better gettering, and octal bases allowing the use of a universal socket design for lots of different tube types.
IMHO The 76 is not the best match in terms of Rp that you could make to the 10K James transformer. I think the 6J5G or 6SN7 would be a better choice. Other types with somewhat to considerably lower Rp would be even better.
5687, 5842/417A are other types I would consider for this application as long as Ip is kept to slightly below 20mA to avoid saturation issues in the transformer.
Note you will get better low frequency extension with tubes with an Rp below 10K when using the 10K tap - you will also probably get less distortion as well. Kind of the point with my harping on the Rp issue..
Edit: kept adding thoughts LOL
I wish I could, as it appears to be a very useful program. However, I use a Mac.
Ah, now I can understand why the 76's Rp of 9.5K isn't optimal with my James transformers.
The 5842/417A seems to be the best option. As I was "google-ing" for more information, I came across the Euridice linestage which appears to be quite popular. However, I have a feeling that it is running the 417A over 20ma. Are there any specific operating points you would recommend? (I am not sure how to adopt the 417A into my present linestage/PS)
Would something like this be reasonable:
An externally hosted image should be here but it was not working when we last tested it.
Out of curiosity, why is the value of the ultrapath cap so low?
Thanks for your help,
Matt
P.S. Anyone wanna trade a pair of 417A's for some 76's
The tube's bias is being set by the cathode resistor's value via the voltage drop across it. Given that the grid is on ground by comparison, this creates the negative bias. The negative bias sets the operating point and the current in the tube depending on the B+ value.
Oh yes, the value of the cap to the cathode sets the LF rolloff point electrically, notwithstanding the LF rolloff of the xfrmr (its limitations) and the saturation due to DC on the primary. LF response too low is not of great value in many circuits, fwiw.
In general terms with this circuit, I find myself asking two questions:
- why do I want or need a 600 ohm output?
- why do I want or need a "volume control" at the input, and on the grid of the tube?
My personal answers are:
I don't want or need 600 ohms at the output, and regardless won't get a proper loading of the transformer circuit - the plate won't see 20kohms - unless there is something like 600 ohms to load the secondary.
Also, the value of the plate load should be adjusted to fit the specific and particular tube and the load line you intend to run it at! Random plate loading values = random results.
So, this brings me to the second question, the input pot. I really really do not want a pot with A) DC voltage across it or B) being part of a DC bias circuit at the grid. IF you used a step attenuator at the input, with DC on the attenuator you'd get "thumps" every time you changed the setting. That should be enough to tell you that it is a bad idea. Not to mention that it does change the grid operating parameters too...
What's the solution?
Imho, the way to go is to put either a fixed resistor to ground or else a fixed attenuator at the input, depending on what your input needs to "see". One could make a good case for a "T" network so that the grid can still see a very high Z while the input can see a Z more like 100kohm or 47kohm for exampe - or if you are working in low Z the input can look like 1kohm for example and the grid can still see 1 megohm... set the input network so that the tube is run in a linear region, and is not overdriven and you get sufficient output on the other end.
Then put your level setting attenuator at the output side.
I'd use a discrete "L" network and a multiposition switch with 2dB steps. No DC present, so all good. Of course you can use a suitable pot, but keep in mind that at 600 ohms or 1kohms there is some current flowing through the pot, so size it appropriately.
Unless you are driving very very long cables and need the low Z you can now select an output transformer with a "closer" ratio, which often means better performance. The attenuator's input Z will be selected to match the desired constant load Z that the transformer wants too see.
Keep in mind here that the current through the tube has to be set such that the step down ratio will result in enough current being available to drive the impedance that you select as the secondary impedance.
IF you build the circuit as shown, and use the 600 ohm output to run a 47kohm (typical high Z input) load, then the tube is not seeing 20kohm at all, and is not operating where it seems like it is... AND the xfmr is essentially unloaded and/or the load changes depending on what you plug it in to - which may or may not be good depending on what the results actually are. (unloaded xfmrs tend to overshoot and ring)
Finally, there is still DC flowing through the primary, so the xfmr should still be the gapped type intended for SE operation. So, maybe a "parafeed" set up offers some economy by getting the DC off the Xfmr primary??
_-_-bear
Oh yes, the value of the cap to the cathode sets the LF rolloff point electrically, notwithstanding the LF rolloff of the xfrmr (its limitations) and the saturation due to DC on the primary. LF response too low is not of great value in many circuits, fwiw.
In general terms with this circuit, I find myself asking two questions:
- why do I want or need a 600 ohm output?
- why do I want or need a "volume control" at the input, and on the grid of the tube?
My personal answers are:
I don't want or need 600 ohms at the output, and regardless won't get a proper loading of the transformer circuit - the plate won't see 20kohms - unless there is something like 600 ohms to load the secondary.
Also, the value of the plate load should be adjusted to fit the specific and particular tube and the load line you intend to run it at! Random plate loading values = random results.
So, this brings me to the second question, the input pot. I really really do not want a pot with A) DC voltage across it or B) being part of a DC bias circuit at the grid. IF you used a step attenuator at the input, with DC on the attenuator you'd get "thumps" every time you changed the setting. That should be enough to tell you that it is a bad idea. Not to mention that it does change the grid operating parameters too...
What's the solution?
Imho, the way to go is to put either a fixed resistor to ground or else a fixed attenuator at the input, depending on what your input needs to "see". One could make a good case for a "T" network so that the grid can still see a very high Z while the input can see a Z more like 100kohm or 47kohm for exampe - or if you are working in low Z the input can look like 1kohm for example and the grid can still see 1 megohm... set the input network so that the tube is run in a linear region, and is not overdriven and you get sufficient output on the other end.
Then put your level setting attenuator at the output side.
I'd use a discrete "L" network and a multiposition switch with 2dB steps. No DC present, so all good. Of course you can use a suitable pot, but keep in mind that at 600 ohms or 1kohms there is some current flowing through the pot, so size it appropriately.
Unless you are driving very very long cables and need the low Z you can now select an output transformer with a "closer" ratio, which often means better performance. The attenuator's input Z will be selected to match the desired constant load Z that the transformer wants too see.
Keep in mind here that the current through the tube has to be set such that the step down ratio will result in enough current being available to drive the impedance that you select as the secondary impedance.
IF you build the circuit as shown, and use the 600 ohm output to run a 47kohm (typical high Z input) load, then the tube is not seeing 20kohm at all, and is not operating where it seems like it is... AND the xfmr is essentially unloaded and/or the load changes depending on what you plug it in to - which may or may not be good depending on what the results actually are. (unloaded xfmrs tend to overshoot and ring)
Finally, there is still DC flowing through the primary, so the xfmr should still be the gapped type intended for SE operation. So, maybe a "parafeed" set up offers some economy by getting the DC off the Xfmr primary??
_-_-bear
bear said:
Sorry to bring this old thread back, but I want to learn more about transformer coupled preamp....
My question is, should I load the output of the transformer with a 1k or 2k resistor so that the load seen by the tube can be determined? (thus I can calculate the tube operating point, current, etc)
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.