So, I've got all the parts for a Quasimodo on the way, and started thinking what I can play around with and snub around here.....that phono preamp I built needs attention.....that Lou Reed C.D......but standing around the oscilloscope the room is filled with guitar amps in various states of functionality, which is where I shall start.Transformer is transformer. On the bench is a simple Fender valve amplifier, but that set a few quandaries.
The power transformer has three secondaries.
High Tension..........300V@ 180mA......... Full Wave Bridge Rectifier...... Capacitor Input Load .......(No questions here, standard setup)
Heater supply...........6.3V@ 2.8A............Unrectified........... Unlikely to ring from its own circuitry, but susceptible to latching on to other transformer oscillations. Also, surely it is through the valves heaters that the most interference it injected into amplifier. Therefore I would assume it is a candidate for some sort of snubbing, maybe another filter tuning be more suitable to purely stop high frequencies if all the other secondaries were clamped to not be able to ring.
Bias & Switching.....55V@ 150mA.........Power surge limiting resistor - 47 Ohm....... Half Wave Rectifier - Cd: 40pF (1 MHz)...... Capacitor Input Load. 70µF.
Now this was the setup that got me wondering. Being a LRC circuit, it is already kind of tuned to be a who-knows-what snubber circuit, and a terrible one at that. But, the question is, if one were to leave the setup as it is, and place the complete snubber with a damping factor zeta = 1, before the limiter resistor, would this resistor detune the circuit? One could place the snubber Cs and Rs after the limiter resistor, and include its resistance into the snubber calculation, would this work? Would having the rectifier come off the top of Rs, Cs somehow change things, or could one simply subtract Rlimiter from the established zeta =1 Rs ?
This standard bias supply has remained unchanged since Fender first started useing fixed bias back in 1955. It is to be found in hunderds of thousand amps, working perfectly acceptably. Some of the newer amps now place the limiter after the rectifier, which I guess is one way to stop it interfering with a snubber circuit. I realise most people would consider it a waste of effort to bother snubbing such a setup at all, but hey, why spend big bucks on the perfect HiFi, just to be able to hear the unadulterated recording of a humming, ticking, intermodulating guitar amplifier.
I would probably change the setup on my own amp. An obvious improvement would be to change to a Full Wave Bridge Rectifier, and maybe use a thermistor as limiter, that effectivly drops out of the circuit once its job is done. And of course - also add a Quasimodo tested snubber - Why? Because I can. Or rather, after I've got my test rig built next week, I will be able to.........
The power transformer has three secondaries.
High Tension..........300V@ 180mA......... Full Wave Bridge Rectifier...... Capacitor Input Load .......(No questions here, standard setup)
Heater supply...........6.3V@ 2.8A............Unrectified........... Unlikely to ring from its own circuitry, but susceptible to latching on to other transformer oscillations. Also, surely it is through the valves heaters that the most interference it injected into amplifier. Therefore I would assume it is a candidate for some sort of snubbing, maybe another filter tuning be more suitable to purely stop high frequencies if all the other secondaries were clamped to not be able to ring.
Bias & Switching.....55V@ 150mA.........Power surge limiting resistor - 47 Ohm....... Half Wave Rectifier - Cd: 40pF (1 MHz)...... Capacitor Input Load. 70µF.
Now this was the setup that got me wondering. Being a LRC circuit, it is already kind of tuned to be a who-knows-what snubber circuit, and a terrible one at that. But, the question is, if one were to leave the setup as it is, and place the complete snubber with a damping factor zeta = 1, before the limiter resistor, would this resistor detune the circuit? One could place the snubber Cs and Rs after the limiter resistor, and include its resistance into the snubber calculation, would this work? Would having the rectifier come off the top of Rs, Cs somehow change things, or could one simply subtract Rlimiter from the established zeta =1 Rs ?
This standard bias supply has remained unchanged since Fender first started useing fixed bias back in 1955. It is to be found in hunderds of thousand amps, working perfectly acceptably. Some of the newer amps now place the limiter after the rectifier, which I guess is one way to stop it interfering with a snubber circuit. I realise most people would consider it a waste of effort to bother snubbing such a setup at all, but hey, why spend big bucks on the perfect HiFi, just to be able to hear the unadulterated recording of a humming, ticking, intermodulating guitar amplifier.
I would probably change the setup on my own amp. An obvious improvement would be to change to a Full Wave Bridge Rectifier, and maybe use a thermistor as limiter, that effectivly drops out of the circuit once its job is done. And of course - also add a Quasimodo tested snubber - Why? Because I can. Or rather, after I've got my test rig built next week, I will be able to.........
I guess you could use Quasimodo to measure the secondary leakage inductance(s), and then put those measured values into one or more SPICE simulations. Because simulations are both controllable and observable (without putting your fingers anywhere near lethal voltages!), maybe you will prefer this approach. Just be sure to accurately model the capacitance-versus-voltage of the rectifier(s), including forward bias. Morgan Jones's article in Linear Audio volume 5 shows why.
I know of SPICEs' existance, but will probably never venture to delve deeper than that. Quite honestly I am more at home pocking around in the hard wired world of four hot EL34's and 700Volts. My 1982 oscilloscope being very hightech next to my 1954 tube volt meter. Being a violin maker from profession, my approach to sound improvement is very much a hands-on, adjust and listen, trial and error method. That does not exclude encompassing proved methods of evaluation, such as your Quassimodo presents. But if you have to live with the amount of so called technical breakthroughs and well meaning ballderdash that our proffesion has to had to suffer in the last four hundered years............ Although Morgans' artical would greatly interest me, I do not subscribe to Linear Audio, although Stuart Yanigers' article about the Cathodyne almost tempted me to. So I will just try.....and observe.
You may want to send a Private Message to diyAudio member peufeu asking for recommendations or suggestions. Peufeu uses a no-math approach; he installs an adjustable (potentiometer) snubber into the final equipment, connects his scope using a highpass filter, and dials the potentiometer until the secondary's ringing disappears. {Peufeu's "bell ringer" , his stimulus that makes the transformer secondary leakage inductance ring , is the rectifier(s) already present in the final equipment}
For heater windings that have no rectifier, one possible way to proceed is to choose a Cx which is comfortably larger than any possible estimate of the total capacitance of all heaters and all heater wiring. Not being a vacuum tube guy myself I don't know this number, but I suspect an upper bound is probably 3000pF. If so then choose Cx = 10,000 pF (10 nF). Use Cx = 10nF in Quasimodo, find a CRC snubber for the heater secondary that gives a well-damped secondary waveform which you approve. Then declare victory. A stimulus may never occur, one that makes this particular bell ring. But if it ever does get stimulated, the CRC snubber will annihilate the ringing, immediately.
For heater windings that have no rectifier, one possible way to proceed is to choose a Cx which is comfortably larger than any possible estimate of the total capacitance of all heaters and all heater wiring. Not being a vacuum tube guy myself I don't know this number, but I suspect an upper bound is probably 3000pF. If so then choose Cx = 10,000 pF (10 nF). Use Cx = 10nF in Quasimodo, find a CRC snubber for the heater secondary that gives a well-damped secondary waveform which you approve. Then declare victory. A stimulus may never occur, one that makes this particular bell ring. But if it ever does get stimulated, the CRC snubber will annihilate the ringing, immediately.
Thanks for the input Mark.
With the heater supply on many tube amps, the supply is not a center taped transformer, and so a virtual earth reference is often formed by running two 100 Ohm resistors across the secondaries, sometimes raising the reference up to another higher voltage than earth. Maybe an alternative would be to just run two caps parallel to these resistors if a true earth reference was used, and be satisfied that these will ground out any H.F. interference that may get through into the supply.
With the heater supply on many tube amps, the supply is not a center taped transformer, and so a virtual earth reference is often formed by running two 100 Ohm resistors across the secondaries, sometimes raising the reference up to another higher voltage than earth. Maybe an alternative would be to just run two caps parallel to these resistors if a true earth reference was used, and be satisfied that these will ground out any H.F. interference that may get through into the supply.
One way to learn the resistor value which gives Zeta=1 in that exact circuit, is to use SPICE.
Another way is to find an electrical engineer who will derive the transfer function of that exact circuit, for you. The transfer function provides an equation which relates Zeta, Ltrafo, Cx, Cs, Rvirtualearth, and Rs. Manipulate the equation so Rs is isolated on the left hand side, plug in your numerical values for Ltrafo, Cx, Cs, Zeta, and Rvirtualearth, boom. Done.
A third way might be to build some sort of supermodified bellringer test-jig which represents that exact circuit, including all the weird items connected to the secondary which might affect damping. Then hope that whatever snubs this arrangement in the test jig, will also snub the (exact same) arrangement in the final circuit.
Another way is to find an electrical engineer who will derive the transfer function of that exact circuit, for you. The transfer function provides an equation which relates Zeta, Ltrafo, Cx, Cs, Rvirtualearth, and Rs. Manipulate the equation so Rs is isolated on the left hand side, plug in your numerical values for Ltrafo, Cx, Cs, Zeta, and Rvirtualearth, boom. Done.
A third way might be to build some sort of supermodified bellringer test-jig which represents that exact circuit, including all the weird items connected to the secondary which might affect damping. Then hope that whatever snubs this arrangement in the test jig, will also snub the (exact same) arrangement in the final circuit.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.