Thanks so much for all your hard work and time you put into these videos! I really get a lot out of watching them. I just downloaded and installed LTSpice today.
Hi Harb, I do have a passion for this stuff. It takes patience to make these inductance measurements but giving the meter time to settle down and now that I have moved the setup to a wooden bench, I am confident I am getting reasonably accurate results quicker (and consistent). It would be nice to have the GR 1632-A bridge as it measures 10 uH to 1111 H. Now, that is one heck of a range... Always good to hear from you.
On another note, measuring the DC Resistance of a winding can magnetize the core. A good transformer will produce no even order harmonics - as your spice results in your next video indicates. If the core is magnetized, a second order harmonic will appear and can even become dominant. Perhaps a good subject for a follow-up video ;-)BTW, I've become addicted to your great videos. Thank you!
It is interesting that you mention that about the (perfect) transformer not producing even order harmonics. I did not know that until now and if there are even order harmonics it is the R or shunt L and C that causes it. I learn a lot from you guys when I post videos. Totally agree about magnetizing the core with even tiny currents. The GR 1632-A manual is online and easy to get from the Bama Boat Anchor web site. It describes my issues perfectly starting on page 8 section 2.11 and talks about demagnetizing the core. Thanks again.
Thanks for the reference to the GenRad manual; there's a lot of good info in there. BTW, Igor S. Popovich wrote a great set of books (3 Volumes) "Audiophile Vacuum Tube Amplifiers" - Volumes 2 & 3 have chapters relating to transformers that are quite good.
A coupe LTSpice tricks. To get power dissipated by a component, hover over the part with your curser, press the alt key, and left click. The display will show power. To get average power hold down the ctrl key with the curser over the node name at the top of the waveform and left click. A box will pop up showing average power. If you want to do some math on the waveform, right click on the node name at the top of the waveform and a box will pop up where you can enter math functions, even using other nodes. Hope this helps and doesn't confuse.
i would probably try stepping through a 5Hz-50kHz signal through each winding and take a look at the output measuring impedance. seems like a frequency dependent thing. from that, i'd see where the values are that they spec them at lie on the graph. have to look at it the same way we look at speakers, i'm afraid.
Good vid, Ray, I do have a couple of comments. tho. 1) The GR 1650A manual recommends that the front panel be grounded ( or the unknown component) to reduce stray capacitance. The fact that the bridge balance is so upset by simply moving the transformer (aka DUT) suggests to me that stray test lead capacitance is adversely affecting your test results. Can you retry, grounded panel, with a piece of shielded line, say mini-coax?Or at least route the test leads to keep them separated by several inches? 2) Also, you may want to try using an external signal source at a higher ptp signal level. I discovered the hard way when trying to build a toroid based impedance transformer for a 30M loop (using MFJ antenna analyzer) that the signal source has to be strong enough to get an accurate reading. So my transformer did not live up to the bench measured 1/1.3 SWR when I tried it with real RF at 100W or more. Ditto when trying to characterize misc transformers with siggen and scope. 3 I presume an audio transformer would be designed to be low q to make it as linear as possible but maybe you could correct me. If I am correct, wouldn't you want to use the CRL selector to L serial. per the manual? Otherwise, the video shows you are modeling with the L parallel and a DQ setting of about 7.5 (on the HiQ ring) which is the manual recommendation for HiQ. Maybe I can learn something here, tho? 4) How about doing frequency sweeps? In an ideal component inductance would not change but we know that transformers are not even close to ideal. Regardless, in any event reactance (and therefor impedance) is going to vary with frequency - even in an ideal component, right? HTH, chuck
Marvelous suggestions. I have not grounded the GR. I will try that. I probably should also move it to my other work bench that has a wooden top as I know the big workbench I use is steel and can be "noisy" and does affect S/N measurements on the HP 8903. I have tried grounding the workbench and it only makes the noise worse. I have used external driving source but not at any higher amplitude. I will try that also. I had not thought about the parallel vs serial measurements. I know the difference but I was not aware that the GR did both and I had a selection. I will take note and see how to change that. Indeed, inductance is going to measure different at different frequencies although we know inductance is not frequency dependent thus my resonate frequency measurement comments. Last night I downloaded the manual for a GR 1632-A inductance bridge. It works from 10 uH to 1111H. I am going to see if I can construct a simpler version of this instrument. Many thanks for the suggestions and conversation. I always make a better instrument after posting and getting suggestions from you guys.
Forgot to comment on (4) I have done frequency sweeps on transformers and even a 6 volt filament transformer with a 120 volt primary is flat from around 10 Hz to 100 KHz - I may be off a little but you see where I am coming from. Frequency sweeps at low levels doesn't seem to prove anything. Of course X and thus the Z will vary with frequency but L does not in smaller inductors like the "calibration" ones I showed. Just glancing at the GR manual I just mentioned, they "bias" the core with a little current. I did not show that in the video but I did put as little as 10 mA through the secondary (8 ohm) winding and that dropped the L to a very low level apparently saturating or nearly saturating the core. I have two huge transformers that will deliver over 1.5 KW of power that are shown in some of my videos that I can saturate the core with 150 mA of current thru the primary. I also tried "biasing" the core with a permanent magnets. Whether the magnet was weak or very strong, it did not make a lot of difference in settling down the transformer and thus the measurements. When I said I tried a lot of things I was not kidding. Your suggestions are excellent and I will be working with them tonight. If I can come up with a true and reliable way to measure these OPT I will definitely make a video to share with the world because as I mentioned, I can not find anything on the Internet that really works for iron core devices. Thanks again.
I did ground the chassis to the safety ground in the AC outlet and I also turned the plug over with no difference with either change. I know on my TV7 tube tester, the polarity of the plug does matter and makes a huge difference. I wasn't thinking when I replied the first time about series and parallel and as you know, Ls is for low-Q (0.02-10) and Lp is for high-Q (1 - infinity). Tonight, during this round of measurements I got a Q=8 in high-Q and Q=9 in low-Q ... close enough. I read through the manual on measuring inductance and I did not have the drive voltage all the way up as it suggested but it was at at least 90%. I don't know if it made any difference but I turned it up fully CW as they suggested. As I suspected, moving the whole measuring setup to a wooden bench allowed the meter to settle a bit faster and I once again get the measurements I showed in the video. Still, suddenly moving the transformer, and I tried it with both of the transformers in the video, will peg the meter and it will slowly return to a proper reading. I don't know why. I measured the entire primary at 28.5H and a Q=14.5 and the secondary at 23 mH. Not perfect per the math but close enough to accept as usable. As for driving it harder (10V @ 1KHz) that was much worse. One volt from an external source was OK but it works best with the internal osc. Driving it externally with higher and lower freq. made it very sluggish and way off. Very hard to find a null, if you can even call it a null. BTW, I do use external osc for small capacitors and that does make it more responsive and a quicker null. Overall, with patience and careful attention, it settles down and eventually give a reasonable reading that I accept as accurate enough. There is a beautiful GR 1632-A on Ebay that I want but I am resisting. I over indulge myself with these instruments but I do enjoy and appreciate them. I may weaken tonight and be the proud owner of it - against my will... :-) I very much appreciate your comments and conversation on this subject.
I measure the inductance with all windings open. Measuring inductance with one winding shorted is a test for leakage inductance and putting a proper load on the secondary allows for measuring impedance of the primary or vise versa. I have learned there is a lot more to this process of measuring core saturation than I initially thought. A DC magnetizing current is part of the process but then a LF AC signal is also part of the process, if I understand things right, or at least a little better. My interests have jumped to making a simple circuit to measure mutual conductance (gm) of a tube. Not just for testing a tube but also for having a standard(s) to calibrate a genuine tube tester. I also picked up a very nice 19" rack today that I want to build another large RF amplifier into. Can't decide which tube to use. Got a nice selection to choose from. Maybe a pair of 833A ? That should look nice.
This is really bearing down on a misunderstood measurement/modeling application. Very nice work! My only question is, in the classic short circuit test you ran the voltage up until reaching rated secondary current was applied and then measured complex impedance parameters. I believe that was to address saturation levels and impacts at max specs. It may apply more to 60 Hz AC than through the audio band but low frequencies dominate in impact. I got most of the way through building a mutual conductance tester that would provide for a course manual curve tracing function. I also wanted to match tubes on it or as you said establish a calibration standard. I'll stay tuned to see what you come up with. Perhaps it will motivate me to dust off my project. Don't see too many 833 amp builds. It would be very cool to see one properly measured and characterized. Have to use the mercury vapor rectifiers to go with the nice size bottles of the 833's. Love that! Are you using those cubic foot size trannies you were showing off in one video? Wow are they huge. Just bought a pait of Dynaco Mk3's and I am going to compare original board with new caps to a VTA Octal driver board upgrade. I did an ST70 driver board Upgrade 10+ years ago with a SDS cap board I still use. I'll probably do that to the Mk3's as well. Going to try a set of Sylvania 6An8's I found. Skeptical of quality that's left out there for these tubes. Keep em glowing!
Mark, Yes indeed, I love these subjects. I had to watch this video again to see what I did. I don't know if you have seen it yet but I built a jig to measure high levels of inductance that actually works better than the GR. My highest regards go to the GR engineers but the GR-1650 is a 1 KHz instrument, even though it has an external input, it works best by-far at 1 KHz and is often times very sluggish at other frequencies (external input) especially if they are LF like 40 Hz or so. Every time I post a video and then learn more, often times from comments from you guys, I wish I had done a better job of the video and included these improvements. Anyway, I am all for the gm project. I think it should be pretty simple. In my case I see test points where I can monitor plate current and grid voltage and vary a bias then introduce a 60 Hz driving signal and do the math. I am pretty sure all this is going to be at peak current and voltage levels and not RMS. I don't think we will have any phase issues but if we do the angles will have to be in radians instead of degrees. Not a problem. I think I have settled on the approach to the 833A amp. It is a medium mu triode so it is not a candidate for GG operation. There are actually a couple of builds on the Internet where good gentlemen have posted their success. They keep it simple and drive the grid with a 50 load off the grid and about -130 volts bias, etc so it will only take 4Kv and a bias supply. My issue is I have a beautiful B&W 852 tank coil I want to use but I am concerned the L values will be incorrect for the 833A. I have to do some careful measurements and calculations before starting. I can always make it a one band amplifier using 1/4 inch copper tubing for a tank circuit but that is not what I want to do. That is the way I did it a long time ago and it works great. The B&W 852 coil is for a single 3-1000Z or a 4CX1000A or even a couple of 3-400Z but I have built amplifiers with all of these tubes and want something else. Steaming-hot and high performing amplifiers and I am a bit concerned ;the 833A will not be as good but I have to try it.
I Watched the follow-on video before posting here. The vector (phasor) analysis is nicely presented. Still pondering specs and DC current saturation component for audio xformers and equivalent ckt params for LTSpice. The test fixture you built is very nice and critical to stabilizing the setup for practical use. I missedunderstood (pun intended) your 833 project as an audio amp. Should still be interesting. I am not a Ham operator but ironically spent 20 years as an Antenna/Systems Engineer in the defense industry. I still enjoy the designs of HAM band amateurs and have worked on designs from HF to mmW freqs. Audio has always been a more practical passion. Keep up the intrepid work and..... Keep em glowing!
When I started on the venture to accurately model audio transformers for LTSpice, I was "horrified" by the lackadaisical attitude that seemed to be everywhere with the other models I found. Everyone used a K factor of something like 0.99997 and usually no leakage inductance or C to ground. Well, after chasing the dragon for a while, and making a bunch of models I figured it out and that is, it is a heck of a lot of trouble to measure all of these parameters, which I worked hard at for a while, with basically no improvement, if any, in the model. So in the end, I understood why such a simple model was, certainly for all I needed, sufficient. Thus, humbled again... I spent all of my working years as a defense contractor working for the military-industrial-complex as President Eisenhower labeled it. My world was in computers and management but I always wanted to be one of you guys designing RF equipment - maybe working for Collins Radio... a ham radio operator's dream? I flip between AF and RF. Two different worlds. One is uH and the other is an even stranger world of iron core inductors. I have been really surprised by the difficulty of measuring vacuum tube output transformers but the method I use with the little jig I built is actually much easier than using the GR 1650-A and I believe just as accurate if not better and actually more repeatable than trying to measure these large inductors at 1 KHz. Very much enjoy our conversations. Got to get to work on the 833A... with, of course, MV rectifiers at the bottom of the cabinet. I can see it all in my head.
I can't for the life of me figure out how you got the tube models and symbols in your LtSpice. I downloaded the tube.lib and tube.slb for koren. There seems to be no way to import or make the files work in LtSpice. the instructions on the site don't seem to apply to LtSpice. I've been working on it all day. Maybe you have different models from a different source? Very frustrating.
I definitely understand the frustration. Often times, for me it is just knowing whether to left-click or right-click. :-) In your case I am going to guess you are not doing the Spice directive function correctly or not at all. Open a file or a new schematic, either one, it doesn't matter which. In the upper right hand side of the LTSpice screen find the .op command and click on it. Type in .lib Koren_tubes.inc and hit enter once. This will give you a floating command of what you just typed in and you need to place it in your schematic anywhere that is convenient and left click at the point you want to place it. You have to do this for all of your vacuum tube drawings. Then in your LTSpice drawing you go over to the Component icon and go to the Misc directory. There you will find the triodes, pentodes, tetrodes. This is one tiny step in a long journey. I hope this helped.
Inductance is measured with the secondary winding "OPEN".Stray Inductance is measured with the secondary winding "SHORTED".Measurement frequency is commonly 100Hz.That GenRad Bridge, driven with enough signal to adequately overcome magnetization flux should provide a good measurement. LCR meters can't do this, your bridge can.Compare with this simpler method (hopefully you can view this JPEG): www.diyaudio.com/forums/attachments/tubes-valves/242835d1317582325-proper-measure-transformer-inductance-indmesurements-jpg
The permeability of the core material is very non-linear - it's low when flux density is low, high when flux density is moderate and lowest when near saturation. Permeability also changes with frequency and DC current levels. A good indicator of a fine Audio Grade Transformer is a smaller change in inductance when measured at 100Hz and at 1KHz. Another good indicator is the shape of the Hysteresis Loop (using an oscilloscope).
Hi Gary, I have tried these methods and they do work for small inductance up to maybe 1 H or so. I use a transformer as the oscillator (60 Hz) so I can ground the center point where the R and L meet and place scope probes across the R and the L. When balanced, it is beautiful on the scope how the amplitude of both signals are equal and they are shifted 90 degrees from each other. On the other hand and as is addressed in General Radio and HP instrument manuals that measure inductance, the larger iron core inductors are a special case. Amazingly, these iron cores are very sensitive to external electrical and magnetic forces and accurate and repeatable measurements of L become a real chore. Overall, with patients, I get and trust values of L for the LTSpice models from the GR 1650-A. I considered purchasing a GR 1632-A but after reading the manual, I realized that the challenges I have with the 1650-A are still present in this dedicated L meter when measuring transformers. Many thanks for your suggestions.
My comment "adequately overcome magnetization flux" should have read, "adequately overcome magnetization current, by a significant margin" - This puts the operating point up into the region where the core's permeability is high.
I sure agree and the sum of the voltages across these coils and series resistors if they are used in measurements, will add up to a good bit more than the source voltage. Very interesting.
I agree and I think that problem has been taken care of in the next video where I do deal with most of the vectors and use the phase shift as part of the measurement.
Interesting, they always talk about the beauty of guitar tube amplifiers in creating even order harmonics, yet your simulation is contrary to that idea
Yes, push pull amplifiers suppress the second harmonic which is usually the highest amplitude harmonic in a single ended amplifier. I am not sure where the idea came from that even order harmonics are good and odd harmonics are bad. I assume someone of musical stature proclaimed it long ago and it stuck. No doubt musicians, and listeners, prefer the over-drive sound of tubes vs transistors. Tubes, when over-driven tend to round off the signal whereas transistor amplifiers have more abrupt changes which contain lots of harmonics and our ears don't like it. It is interesting that around the 1930's, a 15 watt SE amplifier might weigh 75 pounds so they looked for more efficient ways to build amplifiers. Nowadays, we have some fantasy that those amplifiers sounded so good and people want to go back to them. Life is strange sometimes...
ElPaso TubeAmps one word, guitarists. It’s self explanatory. I love how they’ll talk about how “spongey” their tube amp is while pointing to their Marshall jcm 800, which has neither rectifier tubes nor cathode biased. Give that some thought! Oh and their universally beloved Ibanez tube screamer pedal has been shown in spice simulation to only have odd order harmonics! And they all describe it as smooth overdrive, when the mantra is that odd order harmonics are shrill
