PIC 16C773 using it's 12 bit A/D converter with 8x oversampling and two INA122 opamps as a custom two channel ammeter. The INA122's are at +400v as I have a common filament circuit for the 300B's and it's the only place I could measure the separate current through each tube.
Been running for 5 years without a hitch. Could add a relay and shut down if one of the tubes runs away, which would be an incredibly good idea as one always does after about two years.
Been running for 5 years without a hitch. Could add a relay and shut down if one of the tubes runs away, which would be an incredibly good idea as one always does after about two years.Attachments
Still playing with the breadboard and tweaking the programmable regulator design to accomodate a range of devices including low voltage, high current I/O for solid state and hybrid device tracing.
I'm pretty sure I'm going to package this up as an open hardware + open source design and kit, using an Android device for local controls and internet connection. The Arduino is just going to set the measurement ranges and run the measurement cycles under control of the Android, which is connected to and powered by the tube tester over USB.
For a wireless control, the tester could use a WiFi interface to a web app
Last edited:
I noticed that Circuit Cellar no longer has my dsPIC controlled tube amp on their web site, so I will post it or put it on my site.
Digilent has a PIC chip board that is pin AND software compatible with the Arduino Mega. It has many more additional I/O's and runs a 32 bit chip at 80 MHz for blazing speed. You can use the built in bootloader to program it in Arduino scetch, or you can blow away the bootloader and use the regular Microchip tools (MPLAB and C). It is available from Digilent, Microchip, and Digikey for about $50. Most Arduino shields fit. I am using one at work, and it has enough power to drive my test radio hardware making GSM ramp by repeated SPI loads to a DAC chip.
Digilent Inc. - Digital Design Engineer's Source
Yes, there is a ChipKit controlled guitar amp being tested on my workbench....far from being ready for prime time yet.
I used a small PIC microcontroller in my hybrid amp to control the output relay.
It holds the relay off for 4 seconds on power up.
The PIC also monitors the output voltage and turns off the relay if a DC event occurrs.
I also use one on an amp test jig and it has saved me a fortune in speakers.
It holds the relay off for 4 seconds on power up.
The PIC also monitors the output voltage and turns off the relay if a DC event occurrs.
I also use one on an amp test jig and it has saved me a fortune in speakers.
The arduino mega (based on the atmega1280) is a much more "useful" uC, though the 128/328's found on the smaller unit's have there place to I suppose.
C# is my forte, But I really don't see the use of a uC in a tube amp, the highest logic level you could ever need would be a "re-purposed" 555 timer to generate PWM for cooling fans
..
I agree if we are talking about HiFi power amps. But if you include a preamp then the uP can handle things like switching between sources, volume, tone and balance controls and to interface with a remote control or even run a web interface.
If we are talking abuot musical instruent amps then there are lots more settings and yo might want to implement "presets" like on a car radio where you can dial-in the sound you want then press a memory button.
But for HiFi power amps there are no controls, not much to be done.
I designed a mixer with digital volume and tone controls so it could be controlled from a PC by USB. It means I dont have to leave my PC to alter the tone or volume of the mixer !
I managed to extract a very clean sounding 20 watts out of a single 6AS7GA with the sections paralleled in SE mode. I can tell you that the tube will melt if you try this without "digital augmentation". How does it work?
The tube amplifier itself looks pretty normal (board on the right). It isn't quite what the average SE amp builder makes though. The smaller tube is a 6EM7 in a two stage SE triode amp. Its main purpose id to develop about 300 volts P-P of drive for the output stage. The output stage is a cathode follower using both sections of a 6AS7GA, or 5998, or even a 6336A in parallel. Why a cathode follower? The cathode follower is relatively immune to ripple on its plate (high PSRR). The amp was tested with a normal bench power supply, and makes about 8 watts in SE mode with a 175 volt power supply and a 600 ohm OPT. This is limited by the dissipation in the output tube.
Now what if we created a variable power supply and modulated it with the audio signal. Set it up such that the output tube sees a constant voltage across it for all signal voltages. It turns out that the voltage dependent nonlinearities inherent in the triode, goes away and distortion nearly vanishes. This is not new, in fact it was patented by Ross MacDonald in 1957, and was the basis for several succesful small signal designs like the SLCF and the front end of many Tektronix scopes. A thread about it is here:
http://www.diyaudio.com/forums/tube...wer.html?highlight=augmented+cathode+follower
I succesfully applied the modulated cathode follower for output stage use, in that thread. The top tube can be a mosfet, but you are still dissipating a lot of energy as heat. What if the top tube is replaced by a modulated SMPS that tracks the audio signal and holds the voltage across the cathode follower output stage constant, and low...like 75 volts. Agian, this is not new, it is done all the time today in cellular base stations and some handsets to double (or more) the RF power amps efficiency.
I was working on the RF stuff nearly 10 years ago. Why do we care about cellular base station efficiency when they have line power? Because 10 years ago Nextel's electric bill was 1 MILLION dollars a MONTH! What does this have to do with HiFi audio amps? Everything. The power output from an SE amp is limited by the output tube's dissipation. SE amps are usually about 10% efficient. If we double the efficiency, we double the power output for the same dissipation.
The technique is simple, you just need some DSP power to sample the incomming audio signal (I tapped the plate of the driver tube) and control an SMPS such that it's output closely follows the audio signal. Microchip makes some dsPIC's for SMPS use that do all the hard stuff in hardware inside the chip. The software just sets everything up and loops forever. There is ample processing power left for houskeeping like bias control, current monitoring, and even tube condition monitoring.
When the magazine article was published I got a call from the owner of a major guitar amp company. Sadly I explained that it didn't sound too good for guitar since the distortion is very low until all headroom is exhausted and then clipping comes on fast and furious. About a year later I got an email from him showing me a patent application in England that was lifted directly from my original project submission.
Circuit Cellar has removed the original project submission, and they want money to see the published article. I can dig up something and post it if there is any interest.
Attachments
I'd love to see the details of this! I'm specifically interested in grounding and isolation between the audio amp and the micro-controller. Did you use a voltage divider, or transformers (or some other witchcraft) to get the voltages within measurement range?
Thanks!
p.s. My interest in this is for a Tublab SE (which sounds so good it brings forth tears) so it's awesome that you are on this thread!
I have used opto-couplers to isolate micro from amplifier.
I used a PIC and opto-coupler to control the reset pin on an irs2092 classs d amp.
I made PDF's of the schematics and typed up a detailed explanation of how it all works, and why I chose what I did.....only to have VISTA eat my post again! This has been going on for a while and it is getting worse. Vista randomly drops the internet connection. Other computers on the same router work fine. The screen displays "internet explorer can not display the web page" or something like that. The text you typed is gone forever.
I spent last weekend building a new computer. I still have a lot of apps to load. I would take this one to the shooting range, but I am convinced there is absolutely nothing wrong with the hardware. The anomalies in function seem to change with "Windows Updates". So this box will be returned to XP which ran OK on it and it will become my FFT audio analyzer. The old one died. It has been a good DVR and media player too. Vista does OK with that.
Either way, here are the schematics. I'll try again with the explanation tomorrow night. It is almost 11PM here now.....good night.
Thanks for the schematics! This is exactly what I was looking for!
In my opinion, Vista was never a good OS. Windows 7 is better, if you have that choice. Of course I still have a couple of XP machines that will likely never be upgraded. Windows 8 looks like it'll be Vista part 2, so I wouldn't hold my breath on that.
Thanks again!
In my opinion, Vista was never a good OS. Windows 7 is better, if you have that choice. Of course I still have a couple of XP machines that will likely never be upgraded. Windows 8 looks like it'll be Vista part 2, so I wouldn't hold my breath on that.
Thanks again!
I'll put up an explanation of how it all works when I have a chance. Got home from work late last night. The local TV is doing their best to stir up a hurricane panic since there is one approaching and another possibility right behind. My guess is that Isaac will go up the west coast of Florida and make life rather soggy for the "occupy" protesters converging on Tampa. If the threat looks real though, the grocery store shelves will be empty by the weekend, so I will go tonight.
The new machine is already built and running 7 pro. Mid level core i5 mild overclock with SSD for boot and apps, 5TB for storage and 16GB of ram. That should last me for a few years. Loading all my software will take a while. I need Windows for Cakewalk Sonar and they are already reccomending W8. After I listened to them and loaded my current machine with Vista Ultimate early on......lets just say I won't be an early adopter of any MS stuff any more.
Vista was reasonably stable for the first year or so, but became more and more schizophrenic as I loaded more and more stuff on it. Now it will fail to load one or more important driver on boot up quite often. Usually it is the sound module or the printer. Lately it will drop off the internet if inactive for an hour or so. EMu has discontinued the high $$$ external sound box I am using, and doesn't have W7 drivers, so I won't buy any more of their (or Creative's) stuff. The old box will become a media player / DVR. I use XP boxes for stand alone stuff.
I work for a large corporation with several thousand Windows PC's. Most of them still run XP. Vista is not allowed on the network, and new boxes run W7. I think I'll just mimic what they do from now on, since there are some smart corporate IT guys that test all that stuff extensively before allowing it on their network.
I'm downloading Windows 8 preview on a computer I'm building for my granddaughter. It is an old ASUS A7V with an Athlon 1.6G processor. I'm building it from stuff I had laying around (only 500mB RAM). It's main purpose is internet access for school work, music, and Karaoke.
She gets to be the guinea pig. I may go ahead and take advantage of the early release sale just to get a couple licenses and actually load it later on my system as well.
She gets to be the guinea pig. I may go ahead and take advantage of the early release sale just to get a couple licenses and actually load it later on my system as well.
Amp details now.....controller next time.
OK, you asked for it. The witchcraft, magic, and smoke containment is all in the modulated boost converters. The rest of the stuff is pretty straight forward. First I will cover the tube amp since it is the easiest to understand. Both the amp and the digital controller share the same ground although this is not obvious in the schematics. In fact during the debug sessions I found I had to run a short fat copper strap between the ground planes in the two boards. The power supply ground is soldered to this strap. This was needed to solve voltage transients and general instability in the boost converters. Everything worked great until the converters approached max voltage. Here there are 450 volt P-P square waves at 1 MHz
See the amp schematic.
The input stage is the small section of a 6EM7. It is a typical gain stage with LED bias and resistive plate load. No processor control is provided, or needed.
The second stage is a grounded cathode amplifier with a CCS load using the big section of a 6EM7. This tube and topology was selected for its ability to swing the plate from 30 volts to 450 volts. I can get over 350 volts P-P ov clean drive from this tube. 6EM7's are highly variable even between tubes of identical construction. There are at least 6 different constructions too. Gain in the first stage, and bias voltage in the second stage vary a lot. Select a pair of tubes with similar first stage gain. The bias for the second stage is microprocessor controlled. R20 and R28 bring bias in from the controller. C12 and C22 form a low pass filter to remove any noise that sneeks in from the boost converter. R8 and R9 (R7, R17) form a voltage divider to reduce the plate voltage to about 50 volts to travel to the logic board. 50 volts or so is safe for the ribbon cable and the SMD parts on the controller. This signal will be split to two circuits. One measures the DC plate voltage for adjusting the bias. The stage operates at a constant current fixed by the CCS, so the bias voltage sets the plate voltage. The other feeds audio into the high speed ADC inside the processor for modulating the boost converter.
The output stage consists of both sections of a 6AS7, 5998A or 6336A paralleled, but with independent bias adjustments controlled by the processor. These tubes are notorious for mismatched sections, so selecting a tube with similar gains and reasonably close bias voltages will allow for the most power output since both sections will share more equally. As with the second stage R26, R34, R30 and R33 brings in the bais and the caps remove noise. R3 and R4 allow for cathode current measurement by the processor. One section is cutoff (max negative bias) while the bias voltage for the other section is determined. Then the process continues for the other section.
The output stage provides all of the current gain for the amp. It operates with a constant DC voltage across it. That voltage depends on the tube in use. Optimum was about 90 volts for the 5998A. The first and second stages provide all of the amps voltage gain.
OK, you asked for it. The witchcraft, magic, and smoke containment is all in the modulated boost converters. The rest of the stuff is pretty straight forward. First I will cover the tube amp since it is the easiest to understand. Both the amp and the digital controller share the same ground although this is not obvious in the schematics. In fact during the debug sessions I found I had to run a short fat copper strap between the ground planes in the two boards. The power supply ground is soldered to this strap. This was needed to solve voltage transients and general instability in the boost converters. Everything worked great until the converters approached max voltage. Here there are 450 volt P-P square waves at 1 MHz
See the amp schematic.
The input stage is the small section of a 6EM7. It is a typical gain stage with LED bias and resistive plate load. No processor control is provided, or needed.
The second stage is a grounded cathode amplifier with a CCS load using the big section of a 6EM7. This tube and topology was selected for its ability to swing the plate from 30 volts to 450 volts. I can get over 350 volts P-P ov clean drive from this tube. 6EM7's are highly variable even between tubes of identical construction. There are at least 6 different constructions too. Gain in the first stage, and bias voltage in the second stage vary a lot. Select a pair of tubes with similar first stage gain. The bias for the second stage is microprocessor controlled. R20 and R28 bring bias in from the controller. C12 and C22 form a low pass filter to remove any noise that sneeks in from the boost converter. R8 and R9 (R7, R17) form a voltage divider to reduce the plate voltage to about 50 volts to travel to the logic board. 50 volts or so is safe for the ribbon cable and the SMD parts on the controller. This signal will be split to two circuits. One measures the DC plate voltage for adjusting the bias. The stage operates at a constant current fixed by the CCS, so the bias voltage sets the plate voltage. The other feeds audio into the high speed ADC inside the processor for modulating the boost converter.
The output stage consists of both sections of a 6AS7, 5998A or 6336A paralleled, but with independent bias adjustments controlled by the processor. These tubes are notorious for mismatched sections, so selecting a tube with similar gains and reasonably close bias voltages will allow for the most power output since both sections will share more equally. As with the second stage R26, R34, R30 and R33 brings in the bais and the caps remove noise. R3 and R4 allow for cathode current measurement by the processor. One section is cutoff (max negative bias) while the bias voltage for the other section is determined. Then the process continues for the other section.
The output stage provides all of the current gain for the amp. It operates with a constant DC voltage across it. That voltage depends on the tube in use. Optimum was about 90 volts for the 5998A. The first and second stages provide all of the amps voltage gain.
OT...
Preview won't load, I get an error 0x00005D which I have found nothing about by Google.
I'll try the RTM download, but I think my problem may be insufficient RAM. I've only got 896Meg and the min looks like 1GB.
I've still got the old drive with XP on it I can install, but it's small.
Preview won't load, I get an error 0x00005D which I have found nothing about by Google.
I'll try the RTM download, but I think my problem may be insufficient RAM. I've only got 896Meg and the min looks like 1GB.
I've still got the old drive with XP on it I can install, but it's small.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.