Cheap high-voltage transistor booster

Started by blackcorvo, December 26, 2013, 12:26:08 AM

Previous topic - Next topic

blackcorvo

Hello!

I just wanted to share here with you guys something I've been messing around with. I'm working on a small push-pull tube amp, and since I happen to have a bunch of MPSA42's, I decided to try a little booster/clean preamp using one of those. Of course, as I usually do with these simple circuits, I ran a test on Livewire just to find a good biasing point and all that good stuff.

Here is the test circuit, showing a gain of something around 70:



And here's another version of the circuit, easier to read:



I've tried it in real life. It works, and sounds okay. I would imagine it probably sounds better at higher voltages, but that may require re-adjusting the bias resistors (R3 and R4).
She/They as of August 2021

PRR

> sounds better at higher voltages

It's got 50:1 negative feedback, it's pretty darn clean as-is. In many situations you will be working this FAR below clipping. I don't think there's any objective "better" in a higher voltage. (Yes, I know things always do sound better when you know you have frightening voltages inside.)

The 1 Meg base resistor seems huge.

MPSA42 has gain at ~~1mA of 90 typical 25 min.

With ~~820 ohms in emitter, the input impedance is 820*90 typical, or like 70K. This loads the stage before it, also the input coupling capacitor.

You would like bias resistor to be "bigger", but 14 times bigger is large.

Another way to look at it: for good bias stability, the "1meg" should be smaller than hFE times the DC emitter resistor. 12K times 90 is 1.08 Meg.

Since the hFE="90" could be 25, you "could" have an MPSA42 with low gain and large voltage drop in the 1Meg. And there's no upper limit on MPSA42 hFE (except a high-volt part can't have huge hFE) so you could have an MPSA42 with very low drop in the 1Meg.

Either way the collector voltage gets well off what your simulator says, and "might" be so far out as to limit your clean output.

Yes, in DIY you can toss any transistors that don't bias-up. OTOH, that can be a pain.

I'd split-the-diff between the 70K input impedance and the 1 Meg value. Say 220K. That won't hurt the input impedance much, and is 4X stiffer against hFE variations.

The 68K+12K bias network is pulling more power than the amp itself. If 150V is expensive, you might want to reconsider that. Offhand I'd change the 12K to 33K and then re-figure the 68K.

I'd think the design bias is very near 1mA. That roughly centers the collector swing. So base should be near 12.6V. With 220K and 33K bias resistors we have about 3V drop due to base current, so 15.5V at C1. 33K and 270K at R3 R4 puts you very near.

Input impedance is near 53K (but can be as low as 20K!). Fine for hi-fi, low-ish for guitar cord circuits. There's a cute trick to bring this up, but Porter's show is on so I'll get to it later.
  • SUPPORTER

PRR

The hFE-sensitivity and low input impedance may be improved. Add a transistor and a bias resistor. While the hi-volt transistor must have low (and typically variable) hFE, with a second transistor you get LOTS of hFE which can be turned into input impedance and bias stability.



input Z = 658K
output Z= 66K
voltage gain = 69 into 220K load
THD 0.6% at 35V pk out
Bias stability:
hFE(Q2)=23 Vout=80
hFE(Q2)=400 Vout=77
  • SUPPORTER

blackcorvo

I didn't really calculate anything for that circuit. I basically just copied a generic transistor preamp, and then changed the values of resistors and such based on the ones I have handy here till it worked to my liking. I also copied the circuit that's simulating the guitar's circuit from Google (I know, I'm lazy and I'm sorry for that).

I unfortunately happen to not have any high-gain PNP's at the moment, but I do have MPSA92's. Maybe there's some way I could use those instead, even tho they're not high-gain devices?
She/They as of August 2021

blackcorvo

I tried out the MPSA92 in software, and it seems to work just as well as a low-voltage hi-gain device for Q1.

Right now, I'm working on a project that will take most of my time and budget, but after it's done I'll try the booster on something.

What comes to mind first is a miniamp with the booster as the preamp and a 6AS5 power amp. If my math is correct, a 110v+110v to 6v @ 5A Power Transformer would work just fine to power this circuit. 6v for the filaments, half of the primary for the mains, and the other half for the B+ supply.
And since gain is not a problem here, I could add a tone stack after the booster, and I think it would still have more than enough signal to overdrive the power tube.
She/They as of August 2021

mac

#5
You can use a NPN MPSA42 buffer instead of PNP MPSA92.
Something like Joe Davisson headphone amp or Blackfire,



mac

mac@mac-pc:~$ sudo apt install ECC83 EL84

blackcorvo

#6
I was looking back at this topic, and something came to mind. Could I use something like this circuit as the inverter stage in a paraphase circuit?

I think it could come in handy for making small builds where space for tubes is limited, or you just don't have many tubes on hand, or even if you want to make a self-split circuit like the Superfly or the Firefly into a "proper" push-pull amp without having to drill another socket hole on your chassis. Could even have a switch to change from self-split to full push-pull (for tonal purposes).

I'll see about testing it while nobody responds.

[EDIT]
I decided to check the gain of the paraphase inverting stage in guitar amps that use it, by using the gain calculator from AmpBooks.com :

On the Supro Thunderbolt, it has a gain of 81.2, and output impedance of 50.75k.
On the Fender Deluxe 5C3, it has a gain of 80, and output impedance of 50k.

Which is surpridingly close to the values proposed for this MPSA42 circuit.
Am I correct to assume that I can use the same values for the input resistor divider for this stage?
She/They as of August 2021

printer2

You could use a LND150 as a cathodyne PI with little trouble. Since I have a bunch of IRF820's to get rid of I did a fixed bias cathodyne, same circuit in this schematic.

Fred

blackcorvo

Interesting solution! I don't have any high-voltage MOSFETs handy to try that, and I've never had much luch with those. They always seem to be fried when I try to use them.

Which is why I want to try the BJTs.

By the way, I was wondering if the diode trick used on the Vulcan gain stages (https://www.diystompboxes.com/analogalchemy/sch/vulcan.html) could be applied in this circuit. Just curious, really.
She/They as of August 2021

blackcorvo

Basically, what I'm considering is the following:



would it work?
She/They as of August 2021

mac

It works.
The 4m7, diode and 820k form a DC path to gnd. This tiny current turns the diode on a bit.
Tweak the base resistors to change the current.

mac
mac@mac-pc:~$ sudo apt install ECC83 EL84

PRR

> what I'm considering is the following:

Well, I'm confused.
  • SUPPORTER

blackcorvo

Quote from: PRR on November 01, 2018, 09:35:23 PM
Well, I'm confused.

It's the solution for avoiding input overloading used in the Vulcan: https://www.diystompboxes.com/analogalchemy/sch/vulcan.html

Since I'd like to use it as the inverting stage of a paraphase, I wondered if that could help it not get harshly driven by the preamp signal.
She/They as of August 2021

Rob Strand

#13
QuoteIt's the solution for avoiding input overloading used in the Vulcan:
The schematic doesn't quite do that.
The thing about the Vulcan schematic (and Joe's scheme in general) is that the diode kind of linearizes (and softens) things.

If you look at the Vulcan the base current does not go through the diode.  The base current and the
diode current can only come from the 4.7M resistor.  The current is diverted down the diode or allowed to pass through the base.

You can see the pattern here as well,
https://www.diystompboxes.com/analogalchemy/sch/diodeopamp.gif


The PNP circuit might do something but it is has different characteristics.
Send:     . .- .-. - .... / - --- / --. --- .-. -
According to the water analogy of electricity, transistor leakage is caused by holes.

PRR

The paraphase was a slick-trick for up-marketing a one-6F6 radio into a two-6F6 radio.

Given transistors, there are many-many other ways to create push-pull drive to tubes.

A well-designed driver will not crap-out until past the point that the power tubes are slammed flat, so we don't usually worry about input overload (which IS at the user's discretion/indiscretion).
  • SUPPORTER

Rob Strand

Another thing I remembered about Joe's scheme is it does not charge up the input cap when you over load it.  It kind of disconnects under overload making it kind of MOSFET like.
Send:     . .- .-. - .... / - --- / --. --- .-. -
According to the water analogy of electricity, transistor leakage is caused by holes.

blackcorvo

Quote from: PRR on November 02, 2018, 03:31:15 PM
A well-designed driver will not crap-out until past the point that the power tubes are slammed flat, so we don't usually worry about input overload (which IS at the user's discretion/indiscretion).

That's exactly my issue, I'm not a good circuit designer haha! I do things mostly by ear and modifying already-established circuits, but I'm flying blind for the most part. I barely even have anything on transistors when I got my Electronics Technician diploma, and I'm a terrible self-learner, so I don't know what I'm doing 99% of the time.

Case in point: I have the circuit working in a breadboard right now, it's working perfectly! But I was wondering if I could squeeze some more gain from it to match the first stage from a Vox Top Boost circuit. This would allow me to use a couple of 6S52N-V triodes (russian version of the 6CW4 nuvistor) for the rest of the preamp.
My thought process was to increase the B+ voltage on the booster from 150v to the "voxy-er" 275v and maybe reduce the 12k resistor to 10k, so I tried it.
Seems to work fine, no smoke, output still clean as a whistle, but I don't know if this is the best approach, so I'm open for suggestions if any apply.
She/They as of August 2021

antonis

Quote from: blackcorvo on March 16, 2021, 03:17:33 AM
I was wondering if I could squeeze some more gain from it

Lower 680R resistor value..

P.S.
Over 2 years is a good improvement time limit.. :icon_wink:
"I'm getting older while being taught all the time" Solon the Athenian..
"I don't mind  being taught all the time but I do mind a lot getting old" Antonis the Thessalonian..

blackcorvo

Quote from: antonis on March 16, 2021, 07:01:49 AM
Lower 680R resistor value..

Of course!

Quote from: antonis on March 16, 2021, 07:01:49 AM
P.S.
Over 2 years is a good improvement time limit.. :icon_wink:

I mean, it's still sitting here, better put it to use instead of going back to the drawing board. At least until the LND150's I ordered arrive.
She/They as of August 2021

blackcorvo

#19
I tried the following changes to the boost circuit, and I'm quite pleased with them:

- Changed R2 (12k) for 33k;
- Changed R4 (68k) for 100k;
- Changed C3 (22u) for 10u;

With a 470p capacitor for C2, it works rather well as a "top boost" first stage at 275V B+.
I've also added an input circuit similar to what Runoffgroove used for their Britannia circuit (1M and 33k resistors, and LED clipper, but with a bicolor LED - the 2-legged kind).

I'm liking the results.

---

[UPDATE]

- Changed R4 (100k) for 150k;
- Changed R6 (820k) for 1M;
- Changed Q1 (BC327) for BC557 (hFE 257);
- Changed C3 back to 22u;

Better headroom, basically no noise with the guitar turned down.

Voltages (250v B+) :

Q1
E: 0v
B: 13.6v
C: 25.6v

Q2
E: 25.1v
B: 25.6v
C: 121v

Just for the heck of it, I also experimented with a MP25A germanium transistor for Q1. It's slightly noisier than the BC557 with the guitar's volume down, but it's unnoticeable while playing. The gain is lower but not by much, it's a negligible difference IMO.

Voltages (B+ 250v) :

Q1
E: 0v
B: 16.75v
C: 28.5v

Q2
E: 28.2v
B: 28.5v
C: 104.2v


That's all for now!
She/They as of August 2021