Hi all,
I've built a few push-pull amps over the years... mostly Marshall style clones and what not. I've had parts for a KT88 based Single Ended amp based on the AX84 sites 20W SEL sitting around for quite some time. Not sure what's kept me for just diving in on this project so far but it looks like I'll get started next weekend.
Well, I've read that KT88s should not be run with the typical high voltages on the screen (that we normally see with 6L6/EL34) without inviting excessive screen dissipation. I've read of a few different approaches involving regulation or larger-than-normal screen resistors to reign the voltage in - the latter of which results in a sagging screen supply, both require dissipating significant power as heat. I'm wondering if I could do something like the image below: 2 parallel rectifiers, one leading to a cap input supply for the expected filtered DC B+ voltage; and a choke input supply for a reduced screen supply (about 250vDC I estimate) that is still low impedance.
I already have the transformer so I'm working within the constraints of a Hammond 270EX. If someone has some experience to tell me that this idea is a non-starter (or that it would work) I'd appreciate it so I don't have to risk breaking something.
Thanks,
Brian
I've built a few push-pull amps over the years... mostly Marshall style clones and what not. I've had parts for a KT88 based Single Ended amp based on the AX84 sites 20W SEL sitting around for quite some time. Not sure what's kept me for just diving in on this project so far but it looks like I'll get started next weekend.
Well, I've read that KT88s should not be run with the typical high voltages on the screen (that we normally see with 6L6/EL34) without inviting excessive screen dissipation. I've read of a few different approaches involving regulation or larger-than-normal screen resistors to reign the voltage in - the latter of which results in a sagging screen supply, both require dissipating significant power as heat. I'm wondering if I could do something like the image below: 2 parallel rectifiers, one leading to a cap input supply for the expected filtered DC B+ voltage; and a choke input supply for a reduced screen supply (about 250vDC I estimate) that is still low impedance.
An externally hosted image should be here but it was not working when we last tested it.
I already have the transformer so I'm working within the constraints of a Hammond 270EX. If someone has some experience to tell me that this idea is a non-starter (or that it would work) I'd appreciate it so I don't have to risk breaking something.
Thanks,
Brian
When designing power supplies for a screen voltage of about 1/2 the anode voltage there was a "typical" circuit used over and over.
Here is one example of that circuit (see the power supply):
http://www.ozvalveamps.org/playmaster/pm117cct60w2.gif
Here is one example of that circuit (see the power supply):
http://www.ozvalveamps.org/playmaster/pm117cct60w2.gif
Wow, thanks. I've never seen that aproach before. I'm embarrassed to say that i don't get what's holding HT2 at 190vDC... but it looks to fill the requirements. Low impedances, no dividers/droppers and low parts count. The only problem for me would be that it requires a ground referened FWB which will essentially double my plate voltage. The KT88 could handle it but my OT only has taps up to 6k.
Running a cap input and choke input supply from the same secondary is fine. You could even save a diode by running the cap input section via a single diode from the junction of the other two diodes and the choke.
Just be careful about grounding, so one supply doesn't inject buzz into the other. Don't directly ground the secondary CT - take it to the caps negative. Then ground the negative of the PSU output.
Just be careful about grounding, so one supply doesn't inject buzz into the other. Don't directly ground the secondary CT - take it to the caps negative. Then ground the negative of the PSU output.
Matt.B.H.,
You're quite right, I guess it was later than I thought. Below is the correction (upper schematic) and a modified version based on DF96's feedback regarding a single diode for isolation(lower schematic.
Thanks again, everyone. Any other thoughts? Do I have it right now?
Brian
You're quite right, I guess it was later than I thought. Below is the correction (upper schematic) and a modified version based on DF96's feedback regarding a single diode for isolation(lower schematic.
An externally hosted image should be here but it was not working when we last tested it.
Thanks again, everyone. Any other thoughts? Do I have it right now?
Brian
Since your comments gave me enough confidence to wire this up safely I did so, using the lower "3 diode" schematic. I used a 5H inductor feeding a 100uF cap, the other leg fed a 100uF cap directly... no additional loads.
The Cap input section rose to 450vDC, the choke input was lower...425vDC. Not as much of a difference as I was expecting. Obviously adding some kind of load to each bus will drop the V due to transformer regulation but does anyone think the 'difference' between the supplies will increase as well? I would think that the AC load of the ripple current through the cap (combined w/ the AC impedance of the inductor) would be the dominating factor in the voltage drop of a choke input supply, but I'm thinking I'm wrong on something.
Any thoughts before I start thinking about adding a load to the circuit?
Brian
The Cap input section rose to 450vDC, the choke input was lower...425vDC. Not as much of a difference as I was expecting. Obviously adding some kind of load to each bus will drop the V due to transformer regulation but does anyone think the 'difference' between the supplies will increase as well? I would think that the AC load of the ripple current through the cap (combined w/ the AC impedance of the inductor) would be the dominating factor in the voltage drop of a choke input supply, but I'm thinking I'm wrong on something.
Any thoughts before I start thinking about adding a load to the circuit?
Brian
Ok,
I put a resistive load on each rail of 100k and added another 100k in parallel with the first one 2X. At a 100k load the voltages dropped from 450/425 to 445/390; at 50k they dropped to 440/365; at 33k they dropped to 436/350... about 13mA and 10mA of current respectively. So yes, it appears the choke input rail drops faster. When plotted the choke input looks to be leveling off as the current goes up where as the cap input has a linear change in voltage.
So, since the original task was to find a better way to drop the screens without active regulation or resistive dropping, the question now becomes: If the screen voltage is dependent on current, is it really out performing a resistor? Looks like a 5k resistor would drop about 50vDC at the same 10mA load...
I guess I need a data point at the max screen current to see what happens there. Anyone know what a KT88 screen will draw when overdriven? I've heard it can approach the plate current so do I need a data point at 100ma of screen current? Or is 50mA a reasonable place to stop?
I put a resistive load on each rail of 100k and added another 100k in parallel with the first one 2X. At a 100k load the voltages dropped from 450/425 to 445/390; at 50k they dropped to 440/365; at 33k they dropped to 436/350... about 13mA and 10mA of current respectively. So yes, it appears the choke input rail drops faster. When plotted the choke input looks to be leveling off as the current goes up where as the cap input has a linear change in voltage.
So, since the original task was to find a better way to drop the screens without active regulation or resistive dropping, the question now becomes: If the screen voltage is dependent on current, is it really out performing a resistor? Looks like a 5k resistor would drop about 50vDC at the same 10mA load...
I guess I need a data point at the max screen current to see what happens there. Anyone know what a KT88 screen will draw when overdriven? I've heard it can approach the plate current so do I need a data point at 100ma of screen current? Or is 50mA a reasonable place to stop?
DF96,
Sadly no, I wasn't aware until today. But I get to see the effects for myself.
I have a data point at 61mA per rail: 387/250; and 109ma for the choke supply only of 240vDC. I can see based on the best fit curve excel provides that regulation is 'better' than resistive dropping, particularly between 61 and 109mA. A 5k resistor would bring the rail down to 250v at 50ma as well but drop to 0 before reaching 100ma of current. At 250 volts and 61mA the screen dissipation is already at 15W - at 109ma that jumps to 26W so a series resistor is needed. A 2k resistor would be in order to get the dissipation below 8W.
I don't see the value in this approach any more... a 5k series dropper, followed by a filter cap and a 1k series at the screen grid does as good of a job as the choke supply below 60mA of screen current. I really wish I had some data on what the screen really draws when overdriven but I suppose I can collect my own soon enough!
Brian
Sadly no, I wasn't aware until today. But I get to see the effects for myself.
I have a data point at 61mA per rail: 387/250; and 109ma for the choke supply only of 240vDC. I can see based on the best fit curve excel provides that regulation is 'better' than resistive dropping, particularly between 61 and 109mA. A 5k resistor would bring the rail down to 250v at 50ma as well but drop to 0 before reaching 100ma of current. At 250 volts and 61mA the screen dissipation is already at 15W - at 109ma that jumps to 26W so a series resistor is needed. A 2k resistor would be in order to get the dissipation below 8W.
I don't see the value in this approach any more... a 5k series dropper, followed by a filter cap and a 1k series at the screen grid does as good of a job as the choke supply below 60mA of screen current. I really wish I had some data on what the screen really draws when overdriven but I suppose I can collect my own soon enough!
Brian
Last edited:
Calculate the current required for Critical Inductance for choke input...current below Lcrit, you will just approach typical 1.4 x cap input...
Lcrit = Vdc / Ima
where Lcrit is the minimum (critical) choke inductance, Vdc is the supply DC output voltage, and Ima is the load current in milliamps.
Note: I think I got this right...
Lcrit = Vdc / Ima
where Lcrit is the minimum (critical) choke inductance, Vdc is the supply DC output voltage, and Ima is the load current in milliamps.
Note: I think I got this right...
Last edited:
There is a trick which you could try: convert it to a resonant choke supply. Put a capacitor in parallel with the choke to resonate it at roughly 120Hz. This will significantly reduce the minimum current draw needed. The cap needs to be able to cope with some regular AC current.
The idea is to raise the effective impedance of the choke at 120Hz. The snag is that it reduces its impedance above 120Hz so you will need to add extra RC smoothing stages to get rid of buzz. If only a low current is drawn these need not drop much DC voltage.
The idea is to raise the effective impedance of the choke at 120Hz. The snag is that it reduces its impedance above 120Hz so you will need to add extra RC smoothing stages to get rid of buzz. If only a low current is drawn these need not drop much DC voltage.
Sometime the resonant capacitor across the choke is an RC network. The network can also alleviate transient choke voltages if there is no capacitor to ground on the diode-choke node. The capacitor DC voltage rating needs to accommodate the turn-on condition where one end of the choke is at +VACpk and the other end is still at 0V, and similarly where one end of the choke is at –VACpk, and other end is at peak B+ (ie. +VACpk), which is more onerous. The cap may end up being a physically large metalized plastic type.
Thanks for the help, and the additional information regarding manipulating the resonance of the choke. I think I'm bailing on this direction altogether and will go with either an aggressive resistive dropper or a zener + mosfet solution (either a regulator or an amplified zener to drop the supply by about 200vDC).
Can anyone advise on how to predict the current/ appropriate voltage on the screen of a KT88 for a given load or anode voltage? I'm starting to get my head around load lines but I haven't seen a good tutorial on calcuating screen parameters for pentodes (or pentode mode tetrodes). I also can't find a mutual characteristics curve for the KT88, which is what I hear is needed to do such calculations.
Brian
Can anyone advise on how to predict the current/ appropriate voltage on the screen of a KT88 for a given load or anode voltage? I'm starting to get my head around load lines but I haven't seen a good tutorial on calcuating screen parameters for pentodes (or pentode mode tetrodes). I also can't find a mutual characteristics curve for the KT88, which is what I hear is needed to do such calculations.
Brian
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.