Hi,
My question is about the following article:
http://www.pritchardamps.com/pritchardamps/tech_talk/db_magazine.pdf
There's two circuits that should emulate tube behavior..
but I just can't figure out what the box (rectangle with 4 connections) with AX7 in it does in the circuit???!!!
I mean, how should this be understood?
Anyone has any idea?
I'd really like to simulate and study that circuit.
Thanks
Barre
My question is about the following article:
http://www.pritchardamps.com/pritchardamps/tech_talk/db_magazine.pdf
There's two circuits that should emulate tube behavior..
but I just can't figure out what the box (rectangle with 4 connections) with AX7 in it does in the circuit???!!!
I mean, how should this be understood?
Anyone has any idea?
I'd really like to simulate and study that circuit.
Thanks
Barre
Mr. Pritchard tried real hard, bless him, but when he tried to bring this to market it fell on its face. The consensus was that it really did not sound like tubes. Pritchard got angry and had some negative things to say about the amplifier industry. In what I saw of his guitar amp product it was not a very good implementation - kinda clumsy.
This is not very recent stuff by they way. If I recall it was from maybe 15 years ago. Unless he has resurfaced with it.
He was very proprietary about it, so the rectangle in the drawings is his emulator device. If you wanted to make simething with it, you got the op amps and stuff and his things and wired it up. We did make inquiries, but he did not make them available in general. He was trying to sell them as part of products he would make. When he said he would send technical materials, all we got was what you read, no specific stuff. I got the impression he was more interested in attracting investors in his venture than trying to drum up customers for the devices.
Since every tube type has its own transfer function, there was an emulator device for any given tube he cared to emulate. So the device here was the 12AX7 emulator which he called his AX7. That is what you are seeing.
Note in the drawibngs there were lead labels like K and P. These werre the legs of the device that corresponded to the cathode and plate of a tube. Note the low voltage of the circuits.
He had all the graphs and drawings and stuff, but the proof is in the pudding. The product might have had more of a chance in the hifi area, since guitar amps - his chosen market - run tubes well beyond their curves and polite behavior. Hifi sticks to the linear parts of the ccurve.
This is not very recent stuff by they way. If I recall it was from maybe 15 years ago. Unless he has resurfaced with it.
He was very proprietary about it, so the rectangle in the drawings is his emulator device. If you wanted to make simething with it, you got the op amps and stuff and his things and wired it up. We did make inquiries, but he did not make them available in general. He was trying to sell them as part of products he would make. When he said he would send technical materials, all we got was what you read, no specific stuff. I got the impression he was more interested in attracting investors in his venture than trying to drum up customers for the devices.
Since every tube type has its own transfer function, there was an emulator device for any given tube he cared to emulate. So the device here was the 12AX7 emulator which he called his AX7. That is what you are seeing.
Note in the drawibngs there were lead labels like K and P. These werre the legs of the device that corresponded to the cathode and plate of a tube. Note the low voltage of the circuits.
He had all the graphs and drawings and stuff, but the proof is in the pudding. The product might have had more of a chance in the hifi area, since guitar amps - his chosen market - run tubes well beyond their curves and polite behavior. Hifi sticks to the linear parts of the ccurve.
A very old thread indeed.
For a few more clues as to what architecture a "tube emulator" might have internally, consider Joe Souza's "Trioderizer" article
The Trioderizer - a solid state triode
where he discusses how to achieve a particular mu to match the triode of interest. Note also he mentions the high voltage Supertex MOSFETs, which would allow direct substitution into tube circuits.
I would also suggest that any production version would include temperature compensation for the FET, such as Dimitri Danyuk has shown on this forum. The opamp configurations shown in Figures 6 and 7 of Pritchard's article may be performing this function.
For a few more clues as to what architecture a "tube emulator" might have internally, consider Joe Souza's "Trioderizer" article
The Trioderizer - a solid state triode
where he discusses how to achieve a particular mu to match the triode of interest. Note also he mentions the high voltage Supertex MOSFETs, which would allow direct substitution into tube circuits.
I would also suggest that any production version would include temperature compensation for the FET, such as Dimitri Danyuk has shown on this forum. The opamp configurations shown in Figures 6 and 7 of Pritchard's article may be performing this function.
Or you may just run a patent search and read all his patents very, very thoroughly. The explain the stuff in a very good detail, but in bits and parts so it takes a while to open up.
Pritchard's circuits are not similar to trioderizer.
The rectangular box is an element of "non-linear resistance" and Pritchard presents several designs for such. In its "simplest" form it's a ladder of diodes and resistors, developing a non-linear transfer curve similar to that of a triode going to cutoff or to saturation - depending on the purpose where the non-linear element is needed. Later revisions of the circuit eliminated component count by adding in transistors as active elements. Think of the design as diode clipping with the difference that it operates "backwards": The network connects from output to DC potential at the terminal that is equivalent of "anode", hence the network is forward biased and stops conducting when the output signal rises above an amplitude at which the no longer is forward biasing voltage potential across the non-linear element and "anode". This cutoff is rather smooth, as it is shaped by the diode+resistor ladder to be such. A similar network but with different transfer characteristics handles the other half wave.
Anyway, it's all explained by the patents... which also go further, describing the power amp designs that are quite different beast than the XGPA triode idea.
Yes, with proper design the transfer curves are similar to triode tubes so the article isn't **:ing on that. Does it sound like tubes? ...Define what that means anyway. IMO, there is no such thing as single tube sound; it's more like the sound of the overall design. Pritchard is very open that his earliest designs didn't really sound all that good but he kept improving them with help of a crew of guitarists willing to test his products and give him feedback. Today his amps seem to have a rather legendary boutique status and many people owning them hold them in very high regard.
Pritchard's circuits are not similar to trioderizer.
The rectangular box is an element of "non-linear resistance" and Pritchard presents several designs for such. In its "simplest" form it's a ladder of diodes and resistors, developing a non-linear transfer curve similar to that of a triode going to cutoff or to saturation - depending on the purpose where the non-linear element is needed. Later revisions of the circuit eliminated component count by adding in transistors as active elements. Think of the design as diode clipping with the difference that it operates "backwards": The network connects from output to DC potential at the terminal that is equivalent of "anode", hence the network is forward biased and stops conducting when the output signal rises above an amplitude at which the no longer is forward biasing voltage potential across the non-linear element and "anode". This cutoff is rather smooth, as it is shaped by the diode+resistor ladder to be such. A similar network but with different transfer characteristics handles the other half wave.
Anyway, it's all explained by the patents... which also go further, describing the power amp designs that are quite different beast than the XGPA triode idea.
Yes, with proper design the transfer curves are similar to triode tubes so the article isn't **:ing on that. Does it sound like tubes? ...Define what that means anyway. IMO, there is no such thing as single tube sound; it's more like the sound of the overall design. Pritchard is very open that his earliest designs didn't really sound all that good but he kept improving them with help of a crew of guitarists willing to test his products and give him feedback. Today his amps seem to have a rather legendary boutique status and many people owning them hold them in very high regard.
Last edited:
thanks Teemuk. I will put those patents on my reading wish list.
In the 70s we needed a nonlinear transfer function to linearize the photodetectors in our two-color pyrometers. We were looking at the blades inside a running jet engine, and to get meaningful data we needed about 200kHz bandwidth so that we could see the temperature profile changing as the blade swept past the borescope optics. The log amps which had been used previously were not fast enough, eventually we used an opamp based circuit with a ladder of diodes and resistors in the feedback path, to obtain a piecewise-linear approximation of the transfer function. Then the partially-linearized signal went to a fast A-to-D, and the data from that to a temperature lookup table for final linearization.
One of our alternative circuits used the R/2R ladder of a multiplying DAC in the feedback of the opamp, and the output of the opamp fed the input of the ADC. The ADC data fed back to the DAC, controlling the gain. This worked, but the DAC delays were significant at high gain so this design went back on the shelf. I later came up with a processing method to compensate the DAC settling errors, but by then we had moved on. Anyway, this method would work at audio frequencies, you could achieve the desired response curve by using a lookup table PROM between the ADC and the DAC.
In the 70s we needed a nonlinear transfer function to linearize the photodetectors in our two-color pyrometers. We were looking at the blades inside a running jet engine, and to get meaningful data we needed about 200kHz bandwidth so that we could see the temperature profile changing as the blade swept past the borescope optics. The log amps which had been used previously were not fast enough, eventually we used an opamp based circuit with a ladder of diodes and resistors in the feedback path, to obtain a piecewise-linear approximation of the transfer function. Then the partially-linearized signal went to a fast A-to-D, and the data from that to a temperature lookup table for final linearization.
One of our alternative circuits used the R/2R ladder of a multiplying DAC in the feedback of the opamp, and the output of the opamp fed the input of the ADC. The ADC data fed back to the DAC, controlling the gain. This worked, but the DAC delays were significant at high gain so this design went back on the shelf. I later came up with a processing method to compensate the DAC settling errors, but by then we had moved on. Anyway, this method would work at audio frequencies, you could achieve the desired response curve by using a lookup table PROM between the ADC and the DAC.
Just wanted to add, overall you're most likely interested in US Patent #5,434,536 "Semiconductor Emulation of Vacuum Tubes". It's the most thorough presentation of Pritchard's triode emulation, discussing the non-linear networks and the whole gain stage idea. A lot of other of his patents patents are further refinements and upgrades to the idea, or present a bunch of additional circuits Pritchard also used in his amps. Overall, the aforementioned patent is likely the one that "makes most sense" and helps to understand many of his other patents and some of the non-standard symbol notation he often uses.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.