I do appreciate the input but most of the time my 'technical abilities' are stretched to the limit on modest projects. I'm sticking with the NTC because it's simple and hopefully it works.
Regards,
Dan
Dan, a snubber normally interacts with waveform details that have frequency content well in excess of what was shown in those spectrum plots.
Those plots are for winding-diode current. With a snubber, the diode current spectrum in that low frequency span would not change (unless the diode had a substantial off-state capacitance, or bypass capacitance was added). The winding current would change, but only with respect to much higher frequency constant, and that may well not be discernible in a test measurement due to the resolution of the current probe.
Some of that higher harmonic content in the linked plots may well be due to mains AC waveform distortion prior to the power supply being connected - ie. the main AC voltage waveform may already by 'flat topped'.
Last edited:
Over in the Cheapomodo thread, a member asked a question which has more to do with the very idea of bellringer test jigs like Quasimodo, and less to do with a "cheapo" way to build one out of junkbox parts.That approach certainly can work, and more than one person here has stated their preference to optimize snubbers this way. Instead of a battery powered test jig named Quasimodo, this method uses the actual diodes in the actual power supply, to provide the bell-ringing pulses that create the ringing waveforms seen on the oscilloscope.
I myself have chosen not to follow that approach, for three reasons:
First, I don't especially want to stick my hands into a piece of live equipment, connected to the AC mains, if I don't have to. Why risk a nasty electric shock when there's an alternative called Quasimodo that runs off a 9 volt battery.
Second, many people* like to build audio equipment using "soft recovery diodes" which do not generate very tall or very steep bell-ringing pulses. So the transformer's secondary oscillations are small and difficult to see on an oscilloscope, even with no snubber attached. It's hard to tell whether you've dialled in the desired zeta=1.0 or not when the scope traces are muddy. Quasimodo, by virtue of its 0.006 ohm output impedance, creates enormous ringing amplitude which is so big it is very easy to see, even on an $18 toy oscilloscope.
Third, I prefer to lay out the 2C+1R snubber(s) in a very compact area of the power supply PCB, right next to the connectors for the secondaries. It would be extremely clumsy to install a temporary potentiometer into this tightly packed area, and even clumsier to reach an adult-sized hand into that area in order to twist a potentiometer shaft. I find it much preferable to adjust the snubber on the lab bench where there's all the room in the world. Then when the optimization is complete, I know the final values of Cx, Cs, and Rs. I proceed to lay them out in a small and tight area on the PCB.
*Including me
I myself have chosen not to follow that approach, for three reasons:
First, I don't especially want to stick my hands into a piece of live equipment, connected to the AC mains, if I don't have to. Why risk a nasty electric shock when there's an alternative called Quasimodo that runs off a 9 volt battery.
Second, many people* like to build audio equipment using "soft recovery diodes" which do not generate very tall or very steep bell-ringing pulses. So the transformer's secondary oscillations are small and difficult to see on an oscilloscope, even with no snubber attached. It's hard to tell whether you've dialled in the desired zeta=1.0 or not when the scope traces are muddy. Quasimodo, by virtue of its 0.006 ohm output impedance, creates enormous ringing amplitude which is so big it is very easy to see, even on an $18 toy oscilloscope.
Third, I prefer to lay out the 2C+1R snubber(s) in a very compact area of the power supply PCB, right next to the connectors for the secondaries. It would be extremely clumsy to install a temporary potentiometer into this tightly packed area, and even clumsier to reach an adult-sized hand into that area in order to twist a potentiometer shaft. I find it much preferable to adjust the snubber on the lab bench where there's all the room in the world. Then when the optimization is complete, I know the final values of Cx, Cs, and Rs. I proceed to lay them out in a small and tight area on the PCB.
*Including me
Last edited:
What the 'in-situ' test environment may provide, that a lab bench may not, are the applicable loading conditions on the other windings of the transformer. On the lab bench, those conditions are estimated, and either a short-circuit, or open-circuit is applied, and 'ground' interconnections may or may not be made between windings, and to the core.
Just lashed up a Quasimodo from 90% of the bits from stock. Only things needed were the shottky's, the MOSFET driver, MOSFET and 1k trimmer. Used a chunk of proto-board. Only one missed wire around the 555, at which point it woke up and worked perfectly.
First test a random inductor to confirm operation, and have tested two transformers thus far:
Chinese R-core, 18V + 18V, 40VA. Measured parameters on bridge - leakage 980uH, C 90pF. Rsnub was 100 ohms for 240V primary
500VA toroid (old ILP unit, 35V secondaries, 240V primary) Leakage 10uH, C 1.1nF. Rsnub 18 ohms.
There is some higher frequency stuff knocking about that others have noticed.
Hagerman's analysis (scaled in proportion for different Csnub) gives 120 ohms for the first Tx and 18 ohms for the second. So all is in agreement on a sample of two very different transformers between Quasimodo and Hagerman.
Thanks Mark - a very useful design that gives a nice optimum using only two capacitance values.
First test a random inductor to confirm operation, and have tested two transformers thus far:
Chinese R-core, 18V + 18V, 40VA. Measured parameters on bridge - leakage 980uH, C 90pF. Rsnub was 100 ohms for 240V primary
500VA toroid (old ILP unit, 35V secondaries, 240V primary) Leakage 10uH, C 1.1nF. Rsnub 18 ohms.
There is some higher frequency stuff knocking about that others have noticed.
Hagerman's analysis (scaled in proportion for different Csnub) gives 120 ohms for the first Tx and 18 ohms for the second. So all is in agreement on a sample of two very different transformers between Quasimodo and Hagerman.
Thanks Mark - a very useful design that gives a nice optimum using only two capacitance values.
Congratulations sawyers! Glad it worked for you. I myself was not able to measure leakage inductance even remotely accurately without Quasimodo. At the time I thought the issue was (a) leakage inductance varies with frequency, i.e., power transformer cores are not linear at frequencies above 10 kHz; and (b) the inductance value I wanted/needed, was L at the resonant frequency of the secondary RLC circuit. Unfortunately I did not know what that frequency was!
Pleased to hear that your bridge measurements of transformer leakage inductance, are trouble free. Morgan Jones (author of Valve Amplifiers) hooked up several transformers to his bridge, and got a messy collection of confusing curves: article in Linear Audio. He suggested modeling the secondary with a six lumped element model (!). I tried to do that with a couple of transformers but didn't have a lot of success.
Since that time I discovered the "Quasimodo ExtraLight" approach which finds an optimum snubber using only a function generator, an oscilloscope, a series resistor, the transformer, and the snubber-to-be-optimized. No PCB, no battery, no oscillator, no gate driver, no MOSFET. diyAudio search will find it for you.
Pleased to hear that your bridge measurements of transformer leakage inductance, are trouble free. Morgan Jones (author of Valve Amplifiers) hooked up several transformers to his bridge, and got a messy collection of confusing curves: article in Linear Audio. He suggested modeling the secondary with a six lumped element model (!). I tried to do that with a couple of transformers but didn't have a lot of success.
Since that time I discovered the "Quasimodo ExtraLight" approach which finds an optimum snubber using only a function generator, an oscilloscope, a series resistor, the transformer, and the snubber-to-be-optimized. No PCB, no battery, no oscillator, no gate driver, no MOSFET. diyAudio search will find it for you.
Last edited:
the "in-situ"test will say it works in real life , our's device are made to works connected to the main and having a load at the secondarie , dont they ??
the Quasimodo looks interresting in a way , is that possible to create such device to test ( ringing ) under working conditions our trafos ??
.
Hi all,
Would somebody in the EU have a Quasimodo rig to lend me? I pay postage of course. I could get PCB made and build the circuit, but sharing is part of the spirit of this place I believe
I have a couple of 400VA 230V/1x18V toroids arriving and I would like to snubber the PSU made with these for a class A amp.
In my parts box I have Schottky 7A/60V (MBR760) and "ultra fast" 8A/600V ).
Will these do, or am I better advised to go shopping different parts?
I have read the article of Mark in linear audio and I think I should be ok even if the the 60V rating might be borderline....
Thanks,
Nic
Would somebody in the EU have a Quasimodo rig to lend me? I pay postage of course. I could get PCB made and build the circuit, but sharing is part of the spirit of this place I believe
I have a couple of 400VA 230V/1x18V toroids arriving and I would like to snubber the PSU made with these for a class A amp.
In my parts box I have Schottky 7A/60V (MBR760) and "ultra fast" 8A/600V ).
Will these do, or am I better advised to go shopping different parts?
I have read the article of Mark in linear audio and I think I should be ok even if the the 60V rating might be borderline....
Thanks,
Nic
I myself don't know whether it's possible to create such device or not. Perhaps you might collaborate with member trobbins, to study the idea. Maybe the two of you will devise something so small and simple that it can fit on a solderless breadboard, like the first generation Quasimodo.
_
Attachments
Remember that snubbers attach to the secondary windings of transformers, and that's where Quasomodo attaches too.
If TransformerA's primary is connected to the AC mains, and also TransformerB's primary is connected to the AC mains, then I would use Quasimodo on each secondary winding of TransformerA. I would optimize a snubber for each secondary winding of TransformerA. I would also use Quasimodo on each secondary winding of TransformerB. I would optimize a snubber for each secondary winding of TransformerB.
Nic, I would be glad to send you all of the electronic parts to implement the low cost version called "Cheapomodo", and you could plug the circuit together on your solderless breadboard / protoboard. Presto you can now test transformers and optimize snubbers! Since the collection of parts is so cheap I will simply give them to you for free, you don't have to send them back to me. Just pay postage from California-USA to you. EUR 2,65 at most. The only thing I ask in return, is that you post a photograph of your protoboard with your assembled cheapomodo, and an oscilloscope photo showing it in operation.
The schematic is below. It's simple and cheap.
_
Attachments
I use an ancient Marconi 0.25% Universal Bridge TF1313A - it is a tube design, but works perfectly. The tricky thing is setting the Q at the null. Because of transformer resistive and magnetic losses the Q for L is really low, about 2-3. So it is an iterative and somewhat tricky thing. But you can easily measure down to single uH's. For C the Q is much much higher because it is dominated by dielectric losses. Two frequencies choosable, 1kHz and 10kHz.
I actually also measured the capacitance on a GR capacitance bridge I have to verify (since the Fluke DVM was telling me porkies) at a fundamental level, and the Q when measuring winding capacitance on a toroid was 140. The GR and Marconi agree on capacitance which was good confirmation.
I subscribe to Linear Audio, so have read your matrix of experiments on rectifiers - very interesting. Yes - I picked up the 6-parameter curve fit; yeesh!
I toyed with the idea, since I have several different suitable function and plain square wave generators, but thought I'd just do a general Quasimodo fixture (since I had most of the bits already) and not worry about remembering how I hooked everything up next time.
Hi all,
Would somebody in the EU have a Quasimodo rig to lend me? I pay postage of course. I could get PCB made and build the circuit, but sharing is part of the spirit of this place I believe
I have a couple of 400VA 230V/1x18V toroids arriving and I would like to snubber the PSU made with these for a class A amp.
In my parts box I have Schottky 7A/60V (MBR760) and "ultra fast" 8A/600V ).
Will these do, or am I better advised to go shopping different parts?
I have read the article of Mark in linear audio and I think I should be ok even if the the 60V rating might be borderline....
Thanks,
Nic
Should you be lazy member dsolodov have complete kit that just need to be soldered, had a kit from him via Paypal in 7 days back in 2014.
dsolodov post 468: http://www.diyaudio.com/forums/powe...-using-quasimodo-test-jig-47.html#post4080314
Doesn't example XSim combined DATS device or soundcard I/O into REW or ARTA work fine for that task.
Attachments
Last edited:
A man after my own heart! Come to NorCal, let us drink a beer and tell stories full of wild exaggerations. I too prefer simple, single-purpose, small test fixtures that are easy to use and hard to connect wrongly with no guidance from a manual. Twenty four months between uses, the damn thing had better be screamingly obvious and self explanatory. When not in active use, it goes into a Rubbermaid plastic food storage bin, at the back of the closet, under the ski clothes and racquetball gloves.
I don't begrudge the couple or three hours to stuff and solder and check out the new gizmo. It's pleasant to put things together and rewarding to see them work; if I happen to get use out of the end result a second time, that's pure profit. The measurement number is half the fun; the home built measurement apparatus is the other half.
post468 is here
http://www.diyaudio.com/forums/powe...-using-quasimodo-test-jig-24.html#post4080314
post876 points to the wrong post.
http://www.diyaudio.com/forums/powe...-using-quasimodo-test-jig-24.html#post4080314
post876 points to the wrong post.
Mark may wish to update his excellent test-jig paper linked from post#1 to describe where the setup for the test-jig with a test transformer may differ from an actual circuit's operational conditions.
For example, an old style full-wave secondary winding that is diode rectified is typically tested by applying Quasimodo to one winding 'side', and shorting the other winding side. In situ, the other winding side is open-circuit at the time of interest (as that diode is off), but that is when there is no snubber loading the winding. So a pedantic tester may want to appreciate that an iterative test process could be worthwhile.
Another example is that the mains supply presents a non-zero but low impedance to the primary winding. The test process is to short the primary winding. In practise, that assumption would typically cause negligible change to a snubber setting. There have been many studies and surveys of typically found 'mains supply' impedance curves with frequency that indicate that the mains impedance at the frequency of snubber interest is likely to be in the range of 10-50 ohm.
Another example is to connect the commonly earthed points of all the windings, and the core, as that represents the capacitive couplings that exist in situ.
I would love to Mark - the slight difficulty being that I'm 5,500 miles away in Oxford UK!