Interesting hybrid mic pre thread
- Thread starter lassoharp
- Start date
Help Support GroupDIY Audio Forum:
Kingston
Well-known member
I wonder what an active plate load does to sound. "pure" current source, right? On paper it removes all "tubeyness" from the plate curves: your load line is now a straight horizontal line, like with a plate choke. Common sense would dictate it results to very very sharp clipping (one end) or blocking distortion (other end) when one runs out of headroom. Any point in using tubes there anymore, forced absolutely linear?
Tests regarding this are on my todo bench list.
Tests regarding this are on my todo bench list.
Yeah, that type of thinking is like you said - aimed at forcing the tube away from it's characteristic curve. It always sounds good on paper. I'm interested in seeing if it improves the load efficiency - ie would it be better than a choke load.
The series grid resistor with that plate to plate FB is interesting. It looks like it will be leaving added noise when the volume of the first stage is turned down.
The series grid resistor with that plate to plate FB is interesting. It looks like it will be leaving added noise when the volume of the first stage is turned down.
I don't know if it would be better, but certainly different. The big advantage of a choke load is the accumulated magnetic energy that allows twice the voltage swing for a given B+, but a SS ccs would be almost totally exempt of the non-linearities of a real-world inductor.lassoharp said:
Kingston said:
Have YOU ever tried it. Try it then talk about it. Then and only then will you be emitting facts. I have current sourced tubes with triode, pentode, Fet, MOSFet and BJT, and guess what...they all sound different. Now explain that to me on paper. When you can accept the FACT that only the theoretical CS is ideal....
Kingston
Well-known member
analag said:
I said I wonder what it does to sound. I asked what would be the point of doing it. Then I implied I want to test things like that.
Which "emitted fact" got you so riled up? I thought this was the place to ask questions and test things.
"they all sound different" doesn't exactly help either. Hey, it's great that you know what's what, no need to rub it in my face.
Every time I have tried hybridi'fying by mixing in such solid-state linearizing add-ons, I've ended up with a circuit that measured and behaved better "on paper", yet did not at all match the subjective perceived performance of a simpler uncorrected setup.
This may very well be due to some deficiency in my perception, but the difference is so clear (to me) that I have given up on such designs a long time ago.
Jakob E.
This may very well be due to some deficiency in my perception, but the difference is so clear (to me) that I have given up on such designs a long time ago.
Jakob E.
gyraf said:
Depends on what you're going for. Personally I like the increased punch and definition especially in the low end. It can be very interesting when done right. I like "all tube" pieces as well as hybrid devices, they all have a time and a place in my audio reality.
I have a mic pre with LND150 depletion MOSFets sitting on top of 5751's and it I find that I use it a lot. Hmmm.
mjk
Well-known member
Hi,
That's my circuit ;D
There are some good reasons for using the particular combination of single ended tube with active load. In my experience it leaves a small amount of low order distortion that is much more musical than the high order products obtained with push-pull or diffamp circuits, or circuits using global feedback.
It's a whole system design using 2 gain stages and local-only feedback. No attenuators or pads needed. The first stage has a high impedance load on the plate, and a low impedance output from the active load that drives the second stage feedback network.
There is still a little first order nonlinearity but hey, if it's stable with 34 dB gain per stage and minimal high order distortion, what's the problem with that? It sounds great at high gain settings; open and effortless, like it's not working hard.
The second stage is pretty much like the first, but allows it's triode plate to see some of the output load reflected through the output transformer. The output transformer ratio is variable to allow different loads and levels to be driven. Some triode character can be produced by driving a low impedance e.g. 600 ohm output load on it's 150 ohm or 600 ohm output impedance setting, or drive it cleanly by selecting the 35 ohm output 8)
It doesn't suddenly clip. The local feedback results in over +36dBu 1st stage input capability at low gain settings and about 150V P-P input capability to the second stage. When you do overdrive a stage, there is no blocking, but a rounding of one side of the sine wave as the stage is overdriven, followed by squaring of the opposite side, then hard clipping. The result is increasing f2 distortion followed by increasing high order products as the stage is driven further into saturation. This behavior is controlled by selecting op point and circuit parameters. It's still a triode and works against it's own internal resistance.
Sonically, it's neutral but not clinical. One listener said "clean and warm" FWIW. It doesn't add a lot of tubyness but that wasn't the goal. The goal was a mic amp that has good gain (up to 75 dB with step-up input ratio) huge headroom (up to +34 dBu input capability) and acceptable noise performance for most mics. At the 1:2 input ratio the EIN at full gain is -125 dBu and the maximum input level at low gain is +28 dBu. I probably increase a little on the noise floor at lower gain, but it still gives me ~150 dB dynamic range. With the DPA high voltage mics it sounds amazing (according to some folks who use these mics in their own recordings). It's the only unit I know of that will handle the entire dynamic range of these mics.
As far as mixing SS and tube characteristics, some circuit techniques like cascode devices and proper device selection and op point are all very important. There is a sonic impact to just about everything and this is no exception. Selection and tuning are sometimes needed. I've found some device combinations that work well together.
I'm currently chasing down some excess power supply noise, narrowed down to radiated EM field from the power toroid (60 Hz plus rectifier switching noise).
I could start a design thread if there's interest. It is intended to be a commercial product but I have no problem discussing the signal path or with others building their own versions. Designing the signal path is only the first 25% of something like this anyway...
Cheers,
Michael
PS I do recommend building and trying. Nothing wrong in speculation; it's just that there's so much more to learn in the doing!
PPS I just figured out where the direct attachment button is (hidden). Here for reference is the abstract of the signal path circuit:
That's my circuit ;D
There are some good reasons for using the particular combination of single ended tube with active load. In my experience it leaves a small amount of low order distortion that is much more musical than the high order products obtained with push-pull or diffamp circuits, or circuits using global feedback.
It's a whole system design using 2 gain stages and local-only feedback. No attenuators or pads needed. The first stage has a high impedance load on the plate, and a low impedance output from the active load that drives the second stage feedback network.
There is still a little first order nonlinearity but hey, if it's stable with 34 dB gain per stage and minimal high order distortion, what's the problem with that? It sounds great at high gain settings; open and effortless, like it's not working hard.
The second stage is pretty much like the first, but allows it's triode plate to see some of the output load reflected through the output transformer. The output transformer ratio is variable to allow different loads and levels to be driven. Some triode character can be produced by driving a low impedance e.g. 600 ohm output load on it's 150 ohm or 600 ohm output impedance setting, or drive it cleanly by selecting the 35 ohm output 8)
It doesn't suddenly clip. The local feedback results in over +36dBu 1st stage input capability at low gain settings and about 150V P-P input capability to the second stage. When you do overdrive a stage, there is no blocking, but a rounding of one side of the sine wave as the stage is overdriven, followed by squaring of the opposite side, then hard clipping. The result is increasing f2 distortion followed by increasing high order products as the stage is driven further into saturation. This behavior is controlled by selecting op point and circuit parameters. It's still a triode and works against it's own internal resistance.
Sonically, it's neutral but not clinical. One listener said "clean and warm" FWIW. It doesn't add a lot of tubyness but that wasn't the goal. The goal was a mic amp that has good gain (up to 75 dB with step-up input ratio) huge headroom (up to +34 dBu input capability) and acceptable noise performance for most mics. At the 1:2 input ratio the EIN at full gain is -125 dBu and the maximum input level at low gain is +28 dBu. I probably increase a little on the noise floor at lower gain, but it still gives me ~150 dB dynamic range. With the DPA high voltage mics it sounds amazing (according to some folks who use these mics in their own recordings). It's the only unit I know of that will handle the entire dynamic range of these mics.
As far as mixing SS and tube characteristics, some circuit techniques like cascode devices and proper device selection and op point are all very important. There is a sonic impact to just about everything and this is no exception. Selection and tuning are sometimes needed. I've found some device combinations that work well together.
I'm currently chasing down some excess power supply noise, narrowed down to radiated EM field from the power toroid (60 Hz plus rectifier switching noise).
I could start a design thread if there's interest. It is intended to be a commercial product but I have no problem discussing the signal path or with others building their own versions. Designing the signal path is only the first 25% of something like this anyway...
Cheers,
Michael
PS I do recommend building and trying. Nothing wrong in speculation; it's just that there's so much more to learn in the doing!
PPS I just figured out where the direct attachment button is (hidden). Here for reference is the abstract of the signal path circuit:
Attachments
Kingston
Well-known member
That's a very curious cathode bias on the stages. Never seen that one before and I don't really get how it works, other than perhaps as a fancy zener acting as fixed bias. Care to enlighten a bit?
One major problem I see in this circuit is current feedback, which yields very low input impedance. IMO, all other benefits the fancy design claims are ruined by this.
mjk
Well-known member
"That's a very curious cathode bias on the stages"
It serves the same purpose as the typical LED bias. The dynamic resistance of an LED is approximately 10 ohms, which is too high relative to the cathode resistance of the D3a working into a load. The diode-connected transistor has a very low dynamic resistance and sets a fixed bias voltage of 1.5 volts for the D3a. Any bias method also contributes to EIN, and this one is very quiet. I didn't try the LED but I would expect too much noise, at least for the first stage. One advantage over a 1n4004 etc. diode is setting the bias voltage to 1.5 volts by using a darlington transistor. I would need a string of 2 or 3 diodes.
"One major problem I see in this circuit is current feedback, which yields very low input impedance. IMO, all other benefits the fancy design claims are ruined by this."
This does result in low impedance, but how does low impedance ruin the benefit? The first stage impedance results in the expected loading for most microphones, and the 2nd stage impedance is easily driven by the low impedance MOSFET source terminal of the active plate load. This really works well and sounds great. The low impedance and local feedback results in good noise rejection.
Cheers,
Michael
It serves the same purpose as the typical LED bias. The dynamic resistance of an LED is approximately 10 ohms, which is too high relative to the cathode resistance of the D3a working into a load. The diode-connected transistor has a very low dynamic resistance and sets a fixed bias voltage of 1.5 volts for the D3a. Any bias method also contributes to EIN, and this one is very quiet. I didn't try the LED but I would expect too much noise, at least for the first stage. One advantage over a 1n4004 etc. diode is setting the bias voltage to 1.5 volts by using a darlington transistor. I would need a string of 2 or 3 diodes.
"One major problem I see in this circuit is current feedback, which yields very low input impedance. IMO, all other benefits the fancy design claims are ruined by this."
This does result in low impedance, but how does low impedance ruin the benefit? The first stage impedance results in the expected loading for most microphones, and the 2nd stage impedance is easily driven by the low impedance MOSFET source terminal of the active plate load. This really works well and sounds great. The low impedance and local feedback results in good noise rejection.
Cheers,
Michael
The values of R2 and R7 not been divulged, it's a tad difficult to estimate. Now, if I understand well, these are pots, so R2 would be reduced only for very high level signals. The input impedance varies with the inverse of the 1st stage gain, so the microphone would see a really improper load only at very low gain settings, so it is an acceptable compromise. Still I don't really like the idea of varying the input impedance as a function of gain.mjk said:
Trident had the same thing in their Series 80; some liked it some didn't. One of the "advantages" of this arrangement is that, contrary to voltage FB stages, noise is lower when the source is disconnected. Makes for better measurements.
Low Z is not a guarantee of low noise; very often ground arrangement becomes more critical.
mjk
Well-known member
I'm sorry, I should have mentioned that it's a simplified schematic.
The 1st stage gain control is actually a 2 section stepped switch that varies the input resistance and feedback resistance to maintain a constant load on the microphone. The prototype is 2500 ohms due to a high-ish DCR input transformer (900 ohms). At the highest gain setting of +33dB, the input resistance is the 900 ohm transformer DCR only, and the feedback resistance is about 112K. At the unity gain setting, the input resistance is the 900 DCR of the transformer plus 1600 ohms additional resistance, and the feedback resistance is about 2500 ohms also. Thus the load on the microphone is constant over the entire range of gain.
The 2nd stage gain control uses a fixed input resistance and a log taper pot for the feedback; about 7K input and 1M feedback. The load on the 1st stage does change a little with gain, from 7K to something like 10K, but the 1st stage output is the low impedance source terminal of the MOSFET. The load line of the 1st tube is constant.
"Low Z is not a guarantee of low noise; very often ground arrangement becomes more critical."
So true as I well know... I use a single star ground but keep the rectifier loops very short in the power supply. I basically follow the practices given by Bill Whitlock and find they work well. A star ground by definition is zero potential because it is the reference. My one tough noise problem has been electromagnetic coupling from the power transformer to the input transformer. With this source decoupled by shielding the power transformer, it's pretty much immune to the quantized AC I get from my off-grid inverter, flourescent lights, etc.
The external wiring is transformer isolated except for the 130V DPA mic input, so it seems to be down to the shielding and balance of the input transformer.
Thanks for the comments!
The 1st stage gain control is actually a 2 section stepped switch that varies the input resistance and feedback resistance to maintain a constant load on the microphone. The prototype is 2500 ohms due to a high-ish DCR input transformer (900 ohms). At the highest gain setting of +33dB, the input resistance is the 900 ohm transformer DCR only, and the feedback resistance is about 112K. At the unity gain setting, the input resistance is the 900 DCR of the transformer plus 1600 ohms additional resistance, and the feedback resistance is about 2500 ohms also. Thus the load on the microphone is constant over the entire range of gain.
The 2nd stage gain control uses a fixed input resistance and a log taper pot for the feedback; about 7K input and 1M feedback. The load on the 1st stage does change a little with gain, from 7K to something like 10K, but the 1st stage output is the low impedance source terminal of the MOSFET. The load line of the 1st tube is constant.
"Low Z is not a guarantee of low noise; very often ground arrangement becomes more critical."
So true as I well know... I use a single star ground but keep the rectifier loops very short in the power supply. I basically follow the practices given by Bill Whitlock and find they work well. A star ground by definition is zero potential because it is the reference. My one tough noise problem has been electromagnetic coupling from the power transformer to the input transformer. With this source decoupled by shielding the power transformer, it's pretty much immune to the quantized AC I get from my off-grid inverter, flourescent lights, etc.
The external wiring is transformer isolated except for the 130V DPA mic input, so it seems to be down to the shielding and balance of the input transformer.
Thanks for the comments!
mjk
Well-known member
analag said:
It was just to show how the signal path works. The actual schematic consists of several sheets with signals going across sheets etc. so I drew a simplified version. I guess I oversimplified it, and I don't have any single diagram that explains the gain structure as well as the text above.
Cheers,
Michael
