Hello,
Between the holidays and also between semesters there will be time to play a little.
You Know over on the bid site there are bunch of used and NOS 12AU7’s and similar. Nothing special about 12AU7’s, there are a bunch of different ones. Often people build DIY audio projects with them; preamplifiers and such. I purchased a few of several different varieties, even a dozen used tubes of various brands. There are no plans to change any attitudes of feelings about the little guy. This is just for fun!
I do have some new and new to me equipment on my bench that I want to get a better feel for.
I want to get an idea about zeroing in on operating points and circuits to minimize noise and distortion. Also to look at THD + N. Then look at the harmonic structure of the THD.
As time permits I want to use op-amps to isolate and actively feedback residual distortion (error only feedback), as in how low can we go in actively reducing distortion?
DT
Between the holidays and also between semesters there will be time to play a little.
You Know over on the bid site there are bunch of used and NOS 12AU7’s and similar. Nothing special about 12AU7’s, there are a bunch of different ones. Often people build DIY audio projects with them; preamplifiers and such. I purchased a few of several different varieties, even a dozen used tubes of various brands. There are no plans to change any attitudes of feelings about the little guy. This is just for fun!
I do have some new and new to me equipment on my bench that I want to get a better feel for.
I want to get an idea about zeroing in on operating points and circuits to minimize noise and distortion. Also to look at THD + N. Then look at the harmonic structure of the THD.
As time permits I want to use op-amps to isolate and actively feedback residual distortion (error only feedback), as in how low can we go in actively reducing distortion?
DT
You can probably go as low as the THD of the opamp itself which means that the sonic signature of the valve will vanish. That means that the valve itself becomes just a fashionable item in the circuit.
Mr. Groucho,
I suspect that you are largely correct.
When the tube is the output device I want to see the FFT plots.
DT
Why not just start with a tube with good linearity to start with and skip the clever engineering dodges?
Look at 6FQ7/6CG7 or 6SN7 as starting points.
Consider as well ECC99 and 6DJ8/6922.
Look at 6FQ7/6CG7 or 6SN7 as starting points.
Consider as well ECC99 and 6DJ8/6922.
Hmmm, department of redundancy department ;-) What are engineering dodges here?
Over the decades, the 6FQ7 has been a common choice of mine, especially as a LTP phase inverter. I also started using a CCS in the tail back in the 80's with excellent results.
Oddly, when I started on a SET design, I needed a 2-stage input/driver that would drive the DHT output stage (45 or 2A3). I opted for a 6FQ7 and spent a fair amount of time with plate curves (and bench testing/measuring) to get operating points that would work well. After not getting the results I wanted (input to output). I switched to a 5814a and after some time working the plate curves a bit, I ended up with much better end-2-end results for THD and overall linearity.
So I think much of how you use the tube in question is in the overall circuit and not just a single stage. Then again, YMMV as I've read where some folks used a LED for cathode bias and a CCS loading on the plate and declared the 12AU7 as bad. I just used basic resistor biasing/loading and direct coupling between stages. Results were really quite good. I've also used the 5814a as a LTP phase inverter with excellent results.
Regards, KM
Over the decades, the 6FQ7 has been a common choice of mine, especially as a LTP phase inverter. I also started using a CCS in the tail back in the 80's with excellent results.
Oddly, when I started on a SET design, I needed a 2-stage input/driver that would drive the DHT output stage (45 or 2A3). I opted for a 6FQ7 and spent a fair amount of time with plate curves (and bench testing/measuring) to get operating points that would work well. After not getting the results I wanted (input to output). I switched to a 5814a and after some time working the plate curves a bit, I ended up with much better end-2-end results for THD and overall linearity.
So I think much of how you use the tube in question is in the overall circuit and not just a single stage. Then again, YMMV as I've read where some folks used a LED for cathode bias and a CCS loading on the plate and declared the 12AU7 as bad. I just used basic resistor biasing/loading and direct coupling between stages. Results were really quite good. I've also used the 5814a as a LTP phase inverter with excellent results.
Regards, KM
Not the typical approach
Hello kevinkr,
I have read many of you posts over the years. Thank you.
I am doing this DIYaudio for hobby fun, just to play with this stuff. I picked out the 12AU7 because there are a lot of varieties of them and they do not cost too much. I also picked out this tube because of its less than stellar performance reputation. I want to watch the effect of adjusting variables.
The effect of adjusting a variable will be more noticeable with a less than stellar performer.
I know that this is not the typical approach.
The first thing I want to look at is noise. The 5814A has a heater that draws 175mA, 25mA more than the standard 12AU7 does at 150mA.
kmaier, The tail of offsetting distortions!
DT
Hello kevinkr,
I have read many of you posts over the years. Thank you.
I am doing this DIYaudio for hobby fun, just to play with this stuff. I picked out the 12AU7 because there are a lot of varieties of them and they do not cost too much. I also picked out this tube because of its less than stellar performance reputation. I want to watch the effect of adjusting variables.
The effect of adjusting a variable will be more noticeable with a less than stellar performer.
I know that this is not the typical approach.
The first thing I want to look at is noise. The 5814A has a heater that draws 175mA, 25mA more than the standard 12AU7 does at 150mA.
kmaier, The tail of offsetting distortions!
DT
Last edited:
Better to see the 12AU7 as two 6C4 in one envelope - because that is what it is.
Under certain circumstances the 12AU7 can give a good approximation to a square law response. If then two are arranged to balance each other (LTP?, SRPP?) the distortion can be quite low. Alternatively, maybe use at lowish signal levels and have low amounts of higher order distortion.
I once threw together a simple cathode-coupled circuit and found that some 12AU7 gave better distortion performance than some 6CG7.
Under certain circumstances the 12AU7 can give a good approximation to a square law response. If then two are arranged to balance each other (LTP?, SRPP?) the distortion can be quite low. Alternatively, maybe use at lowish signal levels and have low amounts of higher order distortion.
I once threw together a simple cathode-coupled circuit and found that some 12AU7 gave better distortion performance than some 6CG7.
If you'r at it, perhaps have a look at the 5963 too.Sort of 12AU7 for computing and very cheap.
Mona
Mona
I've used Mazda Cifte ECC82 (12AU7A) in SRPP and mu-follower circuits as well as cross-coupled differential stages with pretty good results (distortion around -74dBr at 2Vrms out in an SRPP), some Canadian made GE 5814 have also exhibited low levels of distortion in the same circuits. I've also had other 12AU7A (GE, Tungsol [!] and RCA) that in the exact same circuits exhibited much higher distortion. (In some cases as much as 12-14dB worse)
DF96 is dead on the money wrt the relationship of the 6C4 to the 12AU7 which is not a little brother to the 6FQ7.
I will admit I have used 6FQ7 only as SRPP and LTPs so my experience with them is a bit limited, but I got good results with the ones on hand. I have a batch of more suspicious mystery brand 6FQ7s I ought to look at some day.
The 12BH7A is a tube I have generally preferred to the 12AU7A in applications where filament current is not an issue, but I can't vouch for whether or not they are significantly more linear than the 12AU7A, I just liked the way they sounded better. (To me cleaner and more detailed)
DT your comments make sense and it should be an interesting project with plenty of opportunity to observe brand to brand variation which seems rather common with the 12AU7A
DF96 is dead on the money wrt the relationship of the 6C4 to the 12AU7 which is not a little brother to the 6FQ7.
I will admit I have used 6FQ7 only as SRPP and LTPs so my experience with them is a bit limited, but I got good results with the ones on hand. I have a batch of more suspicious mystery brand 6FQ7s I ought to look at some day.
The 12BH7A is a tube I have generally preferred to the 12AU7A in applications where filament current is not an issue, but I can't vouch for whether or not they are significantly more linear than the 12AU7A, I just liked the way they sounded better. (To me cleaner and more detailed)
DT your comments make sense and it should be an interesting project with plenty of opportunity to observe brand to brand variation which seems rather common with the 12AU7A
I have seen that 12AU7 and 6CG7/6FQ7's have a data sheet that appears identical, exception in the middle poit or tab in the heater circuit. Some 6FQ7's have an internal shield wired to pin #9, other not. But the main difference is that 6CG7's have a plate almost twice sized as the 12AU7. They have been designed for horizontal sweep oscillators in a cathode coupled multivibrators and "syncroguide", for tube TV's where large current plate peaks are needed.
Speaking of noise and bandwidth
Making a plan of what tests to do and what tools to use.
I have a cute little QA401 Analyzer that will calculate RMS noise across a user defined bandwidth, which is pretty cool. The AP 2522 is back from service with both A and C weighting filters.
Speaking of noise and bandwidth I am beginning to like the idea of using C-weighting for noise. I do not recall ever being concerned about noise coming from the tweeter. If you measure noise out to a couple 100,000 Khz the noise figure gets really big including frequencies we do not hear.
Exceptions? Additions?
I did order a few 5963’s and 6189’s
DT
Making a plan of what tests to do and what tools to use.
I have a cute little QA401 Analyzer that will calculate RMS noise across a user defined bandwidth, which is pretty cool. The AP 2522 is back from service with both A and C weighting filters.
Speaking of noise and bandwidth I am beginning to like the idea of using C-weighting for noise. I do not recall ever being concerned about noise coming from the tweeter. If you measure noise out to a couple 100,000 Khz the noise figure gets really big including frequencies we do not hear.
Exceptions? Additions?
I did order a few 5963’s and 6189’s
DT
I presume that the balance in the LTP only applies if you are using both phases (i.e. as an actual PI for balanced output or PP driver) is that correct? That is to say if you make a LTP and just take one phase of the output there is no distortion reduction over that of a simple grounded cathode stage.
Correct, I used them that way in some of my early line stage designs. (Maybe even slightly worse than a common cathode stage..)
Valve noise is normally dominated by 1/f noise, so there is not likely to be much difference between C-weighting and unweighted (20-20kHz) measurements. In which case I would encourage you to use unweighted measurements as they are more universally useful to anyone you care to share the numbers with.
No. As we have a AC voltage level at the cathodes of a LTP that equals about half the input signal level, gain - and also distortion - is reduced by the half, compared to a common grounded cathode amp. Best regards!
12AU7 and 6CG7/6FQ7 are not quite the same, and datasheets should show this. Different mu, different gm. My understanding is that a 6CG7 should always have an internal shield, and a 6FQ7 should not have a shield, but manufacturers often label a 6CG7 as 6CG7/6FQ7. You may also sometimes see Mazda 6/30L2 (also known as ECC804) labelled as 6CG7 - but they have half the heater current of a real 6CG7 and similar size anodes to a 12AU7.Osvaldo de Banfield said:
No. LTP balance reduces even-order distortion at both anodes. Perfect balance (i.e. matched triodes, infinite tail) would mean zero even-order distortion, but you don't get that in the real world.mashaffer said:
- Status
- Not open for further replies.