Sure thing Jan here is the thread on ESLDIY,
A TEST JIG FOR FINDING ESL STEP-UP TRANSFORMER PARAMETERS
This the same thread, But this shows using Visual Analyzer that was calibrated to reflect the actual voltages coming out of a Antek AS(N)-1206 in a step up configuration for ESL use,
A TEST JIG FOR FINDING ESL STEP-UP TRANSFORMER PARAMETERS
I used the same technique to build my regulated HVsupply for 0 to 13.8kv.
That project is in these threads,
http://www.diyaudio.com/forums/plan...tor-insulation-mylar-coating.html#post2531218
http://www.diyaudio.com/forums/plan...tor-insulation-mylar-coating.html#post2848194
But here is the detailed version,
Does anyone have schematics of a varible HV power supply
Cheers!!!
jer 🙂
A TEST JIG FOR FINDING ESL STEP-UP TRANSFORMER PARAMETERS
This the same thread, But this shows using Visual Analyzer that was calibrated to reflect the actual voltages coming out of a Antek AS(N)-1206 in a step up configuration for ESL use,
A TEST JIG FOR FINDING ESL STEP-UP TRANSFORMER PARAMETERS
I used the same technique to build my regulated HVsupply for 0 to 13.8kv.
That project is in these threads,
http://www.diyaudio.com/forums/plan...tor-insulation-mylar-coating.html#post2531218
http://www.diyaudio.com/forums/plan...tor-insulation-mylar-coating.html#post2848194
But here is the detailed version,
Does anyone have schematics of a varible HV power supply
Cheers!!!
jer 🙂
Last edited:
The problem is,
Why would anyone want anything but the best?
Each time you complete it, you want better!
So the design goal is I want the best..however how long is it before the best is beaten..😀
So specify the best...the best I can afford at the time based upon the knowledge I have now...trouble is that changes.
It has to look good but not age..with fashion..
Perhaps something that is flexible and can be upgraded easily?
Regards
M. Gregg
Why would anyone want anything but the best?
Each time you complete it, you want better!
So the design goal is I want the best..however how long is it before the best is beaten..😀
So specify the best...the best I can afford at the time based upon the knowledge I have now...trouble is that changes.
It has to look good but not age..with fashion..
Perhaps something that is flexible and can be upgraded easily?
Regards
M. Gregg
Last edited:
I don't always do things the same way every time. The methods and approach depends on what I am trying to accomplish. Sometimes I just sit down with a bunch of parts and play. This usually leads to some fried parts, but sometimes there is great sound to be heard. Other times I rely on my EE training to approach a design from an engineering perspective.
The "pure engineering" approach:
First, I define the finished product. Are there size, weight, and packaging constraints. What is the budget. What are the electrical specs, power output, number and types of inputs and outputs, any heat load or power consumption requirements.... If the amp is for me, I choose realistic distortion figures. I measure every amp I build, but I don't spend time chasing numbers for numbers sake. I prefer to listen to the music, not the test equipment. My favorite amp makes 2 WPC at 5% distortion, and almost 2% at 1 watt. It has gone head to head with some very expensive stuff and fared very well. What is the expected use (HiFi, MI, PA, or all of the above). What is the expected duty cycle, a HiFi that rarely sees full power, or a dance hall PA that eats bass at the edge of clipping for an 8 hour rave. The power supply depends on the duty requirement. Any particular topology requirements, IE the customer wants a DHT amp. Does the customer have any other unique requirements?....
After the product is fully defined, I ask myself if I have a circuit, or combination of circuits in my "bag of tricks" that can be used or adapted for this job. Do I have something already built, or breadboarded, or even something in my "closet of broken dreams" that I can modify or hack into a test amp to verify that my idea for this project will work? If so, I assemble something that is as close as possible to the finished product. Often this may use different transformers than the final device, and usually there will be no power supply, only bench supplies. Then I listen to it, and test its performance to see where it needs improvement. Once I have some confirmation that the intended project will work, then I select the tubes and transformers for the final product and order them. I will use Microsoft Visio to lay out the chassis before building anything to avoid the "it doesn't fit" moments.
Often, I don't have anything in my bag of tricks that will do the job. In this case there are two paths that I take.
If the technology is well known to me (say I need another variant of something I have built before) I go directly to a breadboard of some type. If I believe that there is a high enough probability of success, I will lay out a PC board, make it, populate it, test it, then hack it up and make it work. Again, I listen, and perform instrument tests to verify the performance and improve it where necessary. Then I build the final version.
On a few occasions, I have built the final version from the initial concept. Those were usually variations of something I had, or have done before. One, was a guitar preamp stuck on an SSE board wired as a single channel 50 watt P-P amp. I laid out a PC board for a guitar preamp based on a Marshall design, and wired it to a SSE board with EL34's in it for a screaming lead guitar amp.
If the technology is relatively unknown (mixed tubes and semiconductors, screen drive, cathode follower outputs, etc) I usually attempt simulation before building a breadboard.
Even my SSE amp design was simulated in TubeCad during the development process. TubeCad was very good in it's day, but it is quite limited by today's standards.
About 10 years ago Linear Technology came to Motorola, where I worked, to give a presentation on Switcher Cad which became LT Spice. I went to the presentation and in the middle of dozens of simulations involving linear IC's, SMPS circuits, discrete semiconductor designs, and even simulations at the internal IC level, there it was, a flipping vacuum tube! Yes, somewhere deep inside Linear Technology there was a tube head. The guy giving the presentation didn't know anything, in fact he couldn't explain the circuit (a crystal oscillator using a 6BQ5). He did however give me the entire circuit file including the 6BQ5 model. I decided that the model was pretty crude, didn't know how to deal with grid current, nor correctly model screen current. Vacuum tube models have improved considerably in the past 10 years.
I recently designed and simulated an amp entirely in LT spice. It was a very complex two stage fully differential sweep tube amp that applied drive to both the screen grid, and the control grid of the output tubes. Local feedback was used around the driver and output stage. The simulation predicted a power VS distortion curve that yielded about 100 watts at 1% distortion and 5% distortion at 140 watts. It predicted a plate efficiency at full power of 80%.
I breadboarded the amp and tested it. The performance wasn't too far from the simulated predictions. This surprised me. This is going to be assembled into a 3 RU rack mount chassis for a "jack of all trades" amp that will see sound reinforcement use, and I will play guitar through it. All the parts have been ordered, and received, but one of life's BIG changes has seen my lab disassembled, and it will not be back together for up to a year.
As I said in the beginning, sometimes it's all play and no engineering. What do you do when you find a box full of toroidal power transformers in the dumpster at the warehouse complex. Well most people would either leave them there, or take them to the metal scrap house for $$$$$. Me, well I would take them home and connect them into a tube amp as OPT's. In this case it was Pete Milletts "engineers amp" and yes they worked. Again, I listen, and perform instrument tests to verify the performance. These worked well up to about 50 watts.
Sometimes, it's a unique tube, transformer, or other part. Sometimes it's a wild idea that doesn't lend itself well to modeling. Either way, it's time to play. I have created 4 different vacuum tube breadboarding systems, each an evolution from the previous version. When the lab gets rebuilt, there will be a 5th generation Tubelab breadboarding system.
So how do I design an amp? Well.....it depends......As it says on the Intel Galileo box.....What do you want to build today?
The "pure engineering" approach:
First, I define the finished product. Are there size, weight, and packaging constraints. What is the budget. What are the electrical specs, power output, number and types of inputs and outputs, any heat load or power consumption requirements.... If the amp is for me, I choose realistic distortion figures. I measure every amp I build, but I don't spend time chasing numbers for numbers sake. I prefer to listen to the music, not the test equipment. My favorite amp makes 2 WPC at 5% distortion, and almost 2% at 1 watt. It has gone head to head with some very expensive stuff and fared very well. What is the expected use (HiFi, MI, PA, or all of the above). What is the expected duty cycle, a HiFi that rarely sees full power, or a dance hall PA that eats bass at the edge of clipping for an 8 hour rave. The power supply depends on the duty requirement. Any particular topology requirements, IE the customer wants a DHT amp. Does the customer have any other unique requirements?....
After the product is fully defined, I ask myself if I have a circuit, or combination of circuits in my "bag of tricks" that can be used or adapted for this job. Do I have something already built, or breadboarded, or even something in my "closet of broken dreams" that I can modify or hack into a test amp to verify that my idea for this project will work? If so, I assemble something that is as close as possible to the finished product. Often this may use different transformers than the final device, and usually there will be no power supply, only bench supplies. Then I listen to it, and test its performance to see where it needs improvement. Once I have some confirmation that the intended project will work, then I select the tubes and transformers for the final product and order them. I will use Microsoft Visio to lay out the chassis before building anything to avoid the "it doesn't fit" moments.
Often, I don't have anything in my bag of tricks that will do the job. In this case there are two paths that I take.
If the technology is well known to me (say I need another variant of something I have built before) I go directly to a breadboard of some type. If I believe that there is a high enough probability of success, I will lay out a PC board, make it, populate it, test it, then hack it up and make it work. Again, I listen, and perform instrument tests to verify the performance and improve it where necessary. Then I build the final version.
On a few occasions, I have built the final version from the initial concept. Those were usually variations of something I had, or have done before. One, was a guitar preamp stuck on an SSE board wired as a single channel 50 watt P-P amp. I laid out a PC board for a guitar preamp based on a Marshall design, and wired it to a SSE board with EL34's in it for a screaming lead guitar amp.
If the technology is relatively unknown (mixed tubes and semiconductors, screen drive, cathode follower outputs, etc) I usually attempt simulation before building a breadboard.
Even my SSE amp design was simulated in TubeCad during the development process. TubeCad was very good in it's day, but it is quite limited by today's standards.
About 10 years ago Linear Technology came to Motorola, where I worked, to give a presentation on Switcher Cad which became LT Spice. I went to the presentation and in the middle of dozens of simulations involving linear IC's, SMPS circuits, discrete semiconductor designs, and even simulations at the internal IC level, there it was, a flipping vacuum tube! Yes, somewhere deep inside Linear Technology there was a tube head. The guy giving the presentation didn't know anything, in fact he couldn't explain the circuit (a crystal oscillator using a 6BQ5). He did however give me the entire circuit file including the 6BQ5 model. I decided that the model was pretty crude, didn't know how to deal with grid current, nor correctly model screen current. Vacuum tube models have improved considerably in the past 10 years.
I recently designed and simulated an amp entirely in LT spice. It was a very complex two stage fully differential sweep tube amp that applied drive to both the screen grid, and the control grid of the output tubes. Local feedback was used around the driver and output stage. The simulation predicted a power VS distortion curve that yielded about 100 watts at 1% distortion and 5% distortion at 140 watts. It predicted a plate efficiency at full power of 80%.
I breadboarded the amp and tested it. The performance wasn't too far from the simulated predictions. This surprised me. This is going to be assembled into a 3 RU rack mount chassis for a "jack of all trades" amp that will see sound reinforcement use, and I will play guitar through it. All the parts have been ordered, and received, but one of life's BIG changes has seen my lab disassembled, and it will not be back together for up to a year.
As I said in the beginning, sometimes it's all play and no engineering. What do you do when you find a box full of toroidal power transformers in the dumpster at the warehouse complex. Well most people would either leave them there, or take them to the metal scrap house for $$$$$. Me, well I would take them home and connect them into a tube amp as OPT's. In this case it was Pete Milletts "engineers amp" and yes they worked. Again, I listen, and perform instrument tests to verify the performance. These worked well up to about 50 watts.
Sometimes, it's a unique tube, transformer, or other part. Sometimes it's a wild idea that doesn't lend itself well to modeling. Either way, it's time to play. I have created 4 different vacuum tube breadboarding systems, each an evolution from the previous version. When the lab gets rebuilt, there will be a 5th generation Tubelab breadboarding system.
So how do I design an amp? Well.....it depends......As it says on the Intel Galileo box.....What do you want to build today?
Hi and thank you very much indeed for the very interesting and valuable reply
Speaking of tubes i have only been a little scared by the voltages involved
Not that with solid state is a no risk business ...
I would take profit of your evident knowledge and expertise for asking you some questions:
1) do you start usually working on one channel only or two ?
2) how do you test by listening ? mono ? stereo ? which kind of music do you use ? what aspects of sound do you check ?
3) which speakers do you use for your listening tests ? digital or analog ? which source ?
I believe completely on dual mono approach. I would be religious on that.
But i do not well how to proceed with the listening having only stereo recordings
But the nice thing of the single channel approach is that when you find a single channel working very good two of them will make a really nice stereo system. I am sure of that.
But still i have very little experience on the "mono" world.
Thanks a lot again.
Kindest regards, gino
Most of the time I build a single channel and refine it before building the other one. Keep good notes as to all changes to be sure that your second channel is identical to the first. If I started the project by laying out a PC board for a full stereo amp with a power supply, I will generally populate one channel, refine it, then add the second channel, and finally the power supply.
I have (had) a test bench where my turntable (built in phono stage), CD player, or computer can be switched in as the source. It has the ability to swap channels (left VS right), or sum to a mono output.
I have a TSE, an SSE, a SPP, a class D chip amp, and one of Pete Millett's engineers amps as my reference standards. Any can be played, and the amp under test can be swapped in place of one channel.
There are 3 sets of speakers, a dummy load, and an active subwoofer that can be selected. I usually use the Yamaha NS-10M Studio speakers since they are very revealing and I have been listening to them for about 15 years. The other two sets are home made and relatively new (less than 1 year).
Most of the time I test in stereo with the test amp in place of one channel of the amp that is most similar to what I am testing. I use music that I have listened to for years, since I know where to look for problems. Examples are Candy Dulfer's sax, Joan Baez's guitar and voice, Norah Joan's voice, the dynamics on several different drum tracks, intermodulation on several different rock music tracks where a heavy bass guitar collides with vocals or piano. Beware of poor recordings with IM built in. Metallica's S&M comes to mind. A metal band with a full symphony orchestra could have been a good test source, but the recording sucks.
Keep the breadboard or test amp until the finished product is done, then compare the two to make sure nothing got screwed up in the final design. Much of my early testing is done with lab quality bench power supplies, make sure that the final amp's power supply can handle the transients. I use loud bass and drum testing for this.
Much of the source amp and speaker swapping was done on a patch panel in my old lab. I plan to create a better method for quick A/B tests in the next lab.
I also had the ability to sample the original source at the amp input, and two amp outputs simultaneously and feed this back into the computer's sound card. I set this up for FFT testing of frequency response and distortion, but found that it is possible to record the amp's output into a speaker load with a real music source using a DAW program like Sonar or FL Studio. You can scroll through the resulting wave files and look for anomalies. I plan to expand on this in the future lab.
Hi and thank you very much indeed for the very interesting information
I have already learned how much lay-out in particular can impact performance
One thing are the sim results another another one the actual measurements on a living prototype.
Thanks a lot again. Kind regards, gino
I have already learned how much lay-out in particular can impact performance
One thing are the sim results another another one the actual measurements on a living prototype.
Thanks a lot again. Kind regards, gino
- Status
- Not open for further replies.