Hi Osvaldo,
My thinking was that if I direct couple them the EL84 will have little voltage from anode to cathode because the cathode will need to be higher than the anode of the 12AX7 (I'm trying to configure for 250-300V output). So the 12AX7 dif amp is just amplifying ripple but not setting the bias for the EL84.
Maybe in that case I should move the 12AX7 to before Rs to have more B+, too.
I might be wrong though. It happens all the time 🙂
My thinking was that if I direct couple them the EL84 will have little voltage from anode to cathode because the cathode will need to be higher than the anode of the 12AX7 (I'm trying to configure for 250-300V output). So the 12AX7 dif amp is just amplifying ripple but not setting the bias for the EL84.
Maybe in that case I should move the 12AX7 to before Rs to have more B+, too.
I might be wrong though. It happens all the time 🙂
I apologize by my insistence, but a series regulator is more easy to put to work pretty fine. I made several different of them, always with tubes, form 6CG7 to 6DE7 and ECL82's. And I'm very satisfied with them.
No need to apologize! I built a series regulator recently too and am very satisfied. I'm just looking to learn more about a different style of regulator 🙂
One more attempt with the dif amp before the shunt:
One more attempt with the dif amp before the shunt:
Last edited:
As you said me that understand Spanish, I upload a part of a txt in where I designed the series pass regulator, as an example only.
It has been done in ASCII (Edit.com from DOS 7.10). Then open with word and select MSDOS Encoding to see more or less well.
It has been done in ASCII (Edit.com from DOS 7.10). Then open with word and select MSDOS Encoding to see more or less well.
Try sending an email to jroberts +at + io +dot +com. The magazine material belongs to Joe Roberts and he deserves first dibs on benefiting from his work. If he doesn't sell them anymore then pm me and I'll scan the article for you.
Joe Roberts used to sell the whole collection (16 issues) on CD-R on the bay but it's no longer there. You should try to contact him directly. S.P was an outstanding (pre-internet era) magazine and an absolute must. Try to get them all, you won't be disappointed.
I've encountered what may be a problem here.
The dif amp is driving the EL84s grid and the EL84's grid is biased with a resistor divider from B+. So the load that the dif amp sees will be the divider resistors in parallel, in parallel with the dif amp's load resistor.
To maximize the load and gain from the dif amp, these resistor dividers should be as large as possible. Rg1 max on a EL84 is 1M for cathode bias or 300k for fixed bias. Because we are fixing the bias reference, I think the limit is 300k here. Anyone know if that is correct?
Assuming it is, the maximum value of the resistors in parallel is 300k. We could get about that with a 1M (upper) and a 430k (lower). But that puts the grid 75V above ground, meaning the cathode will need to be even higher. This isn't leaving a lot of room for the voltage drop across the EL84 for 250V output, not to mention the reduced load on the dif amp, and a need for a reference that can take quite a bit of dissipation.
We could lower the voltage at the junction of the divider (270k || 1M = 212k, ~ 53V), but it will be at the cost of a lower load on the dif amp. This could be compensated for with a larger load resistor, but I'd really like to keep some decent current through it. Maybe that's not so important in this application as bandwidth might not be critical.
I could use a different dif amp tube (12AT7?), I suppose. I just have SO MANY 12AX7s so I'd like to try and use them up.
Thoughts, anyone?
The dif amp is driving the EL84s grid and the EL84's grid is biased with a resistor divider from B+. So the load that the dif amp sees will be the divider resistors in parallel, in parallel with the dif amp's load resistor.
To maximize the load and gain from the dif amp, these resistor dividers should be as large as possible. Rg1 max on a EL84 is 1M for cathode bias or 300k for fixed bias. Because we are fixing the bias reference, I think the limit is 300k here. Anyone know if that is correct?
Assuming it is, the maximum value of the resistors in parallel is 300k. We could get about that with a 1M (upper) and a 430k (lower). But that puts the grid 75V above ground, meaning the cathode will need to be even higher. This isn't leaving a lot of room for the voltage drop across the EL84 for 250V output, not to mention the reduced load on the dif amp, and a need for a reference that can take quite a bit of dissipation.
We could lower the voltage at the junction of the divider (270k || 1M = 212k, ~ 53V), but it will be at the cost of a lower load on the dif amp. This could be compensated for with a larger load resistor, but I'd really like to keep some decent current through it. Maybe that's not so important in this application as bandwidth might not be critical.
I could use a different dif amp tube (12AT7?), I suppose. I just have SO MANY 12AX7s so I'd like to try and use them up.
Thoughts, anyone?
Any pair of resistors whose parallel results in a lower value than that is suggested, will work. No restriction exists that one of them must be 1Meg.
In the other hand, the divider can be bootstrapped to make it almost disappear as the plate loading resistor.
The circuit is becoming you crazy, doesn't it? ;-)
In the other hand, the divider can be bootstrapped to make it almost disappear as the plate loading resistor.
The circuit is becoming you crazy, doesn't it? ;-)
Last edited:
OK, actually 375k || 1.5M = 300k and 250 / (375k + 1.5M) x 375k = 50V and 56V zeners are a standard value. 200V across EL84 with a -6V bias is 40mA, leaving 20mA for the load and keeping the max dissipation at spec (12W) with the load disconnected. Hail algebra!
Potential voltage references for 56V:
10 x 5.6V zener (a la Jones)
35 x HLMP6000 || 35 x HLMP6000 (a la SY)
Potential voltage references for 56V:
10 x 5.6V zener (a la Jones)
35 x HLMP6000 || 35 x HLMP6000 (a la SY)
Last edited:
Also, you can use 390KΩ, some few Ω's in excess don't worry. Also, yo can use an ECL84, using the triode as a cathode follower from a neon tube at the cathode's junction.
I looked at neons as voltage references but they all seem very low current. Do you mean using the neon as the signal for the CF and taking the output from the cathode? That would make a good reference, I think.
"Porsupus". I did it several times and really works. The scheme is more or less so: triode plate to Ebb. A neon lamp powered at, say, 100µA is sufficient, from the Ebb supply. Then the junction of neon lamp an the dropping limiter resistor, directly to the grid. Both cathodes joined together and to gnd via an adequately computed resistor (It must carry both section currents). The triode then, will act as a cathode follower from the neon's voltage (I have several of about 60-70V) without any internal resistor the neon only. Note: don't use any capacitor across the neon lamp, it will generate relaxation oscillations.
^^that looks like paralleled triodes to me
with individual grid bias, the resistor at the
cathode you can series a pot to control plate currents.....
with individual grid bias, the resistor at the
cathode you can series a pot to control plate currents.....
to design a shunt regulator, you must start with the load,
what voltage, what current?
then look at your voltage source, what is the terminal voltage,
what is the internal resistance? think Thevenin...
so knowing what your load is, you can then design you shunt reg.
this is how i will do it if that were my project....
imho it would be very difficult design what is in the middle when
you do not know what are at the left or right side of that circuit...
what voltage, what current?
then look at your voltage source, what is the terminal voltage,
what is the internal resistance? think Thevenin...
so knowing what your load is, you can then design you shunt reg.
this is how i will do it if that were my project....
imho it would be very difficult design what is in the middle when
you do not know what are at the left or right side of that circuit...
This is about where I am with parts values:
RE load and voltage source, this would hypothetcially be connected to a phono preamp (class A, low current requirement) and fed by a CLC filter.
I'm calculating with plate curves about 100V in the tail of the dif amp.
RE load and voltage source, this would hypothetcially be connected to a phono preamp (class A, low current requirement) and fed by a CLC filter.
I'm calculating with plate curves about 100V in the tail of the dif amp.
Last edited:
that's better....
the 2500 ohm Rs you can replace with a depletion mosfet
set for 40 - 50mA, and you have a Salas type shunt reg...
the 2500 ohm Rs you can replace with a depletion mosfet
set for 40 - 50mA, and you have a Salas type shunt reg...
imho, overkill, the reason for the shunt reg is to save on parts...
but hey, that is your amp, do not mind me.😀
Thank you AJT. I'll have to read up on the Salas Reg. I have to admit that I get a little lost with solid state stuff sometimes.
And the CLC is definitely overkill 🙂 Maybe I'll try just a cap input filter first and cross my fingers. I'm choosing 9 pin tubes purposefully to try and make this small (probably a EZ81 rectifier for AC to DC).
Something like an Antek 100VA power transformer might be perfect because they have two 6.3V 3A windings. One can by referenced high for the power supply and the second can be doubled and regulated for 12V if this gets attached to a phono preamp (which is where I see this headed).
And the CLC is definitely overkill 🙂 Maybe I'll try just a cap input filter first and cross my fingers. I'm choosing 9 pin tubes purposefully to try and make this small (probably a EZ81 rectifier for AC to DC).
Something like an Antek 100VA power transformer might be perfect because they have two 6.3V 3A windings. One can by referenced high for the power supply and the second can be doubled and regulated for 12V if this gets attached to a phono preamp (which is where I see this headed).
- Status
- Not open for further replies.