BJT/Zener voltage reducer issue

[Cliff Schecht]

I am having issues getting the BJT version of this circuit to work in my 5E8A build. The power supply is probably my most ambitious effort yet because the PT I'm using was the only one I could find that would support 2 5U4's and not cost hundreds of dollars (it cost me $5+shipping on ebay). The HV output is 450V and it has a couple of 6V and a 5V heater tap. Plenty of current available for a 2x6L6GB amp, the transformer weighs in at over 11 lbs. My 5U4's are configured in a hybrid bridge configuration and right now my B+ sits at well over 500V (close to 550V!) without a load. It doesn't drop too much under load either as again this PT is massively overrated. I had to use a different bias supply configuration too because the standard non-adjustable Fender-style bias supply circuit doesn't work with bridge rectifiers. I'm attaching a schematic below.

From what I can tell, the pinout of the BJT in the circuit above is wrong compared to the datasheet (pins 1 and 2 flipped). The circuit calls for a 2N3055 but I need to drop more voltage so I'm using a MJ8505 (800V, 5A transistor). The 2N3055 (http://www.onsemi.com/pub_link/Collateral/2N3055-D.PDF) and the MJ8505 have the same pinouts according to the datasheet which is different then what the PDF specifies.

So I've flipped the leads on my Zeners and bridge rectifier center-tap to both configurations and as far as I can tell neither way works properly. I flip on the standby switch and the voltage goes to about 530V, over the rating of any of my B+ filtering caps. It's been as high as 600V with no load and both rectifiers installed.. So I'm not sure why this circuit isn't working. The diodes I'm using in the reference circuit are 24V, 0.4W 5% Zeners (1N970B) and the current limiting resistor is 100 Ohms (according to MJ8505 datasheet, Vbe at my levels is only about 500mV so I used a 100 Ohm resistor to get about 5mA of current through the Zeners (their Zener voltage is specified at 5mA). My last suspect is that the Zener diodes need to be flipped as well, the way they are in the Zener2.pdf circuit seems backwards..

So what am I doing wrong? Anybody else played with this circuit?

[Cliff Schecht]

I've tried pretty much every polarity I can think of as far as BJT and Zener orientation but as far as I can tell it isn't working, the voltages are still much too high. The 6L6's (Sylvania 6L6WGB) I'm using are tough as nails though, they are taking the over 500V B+ like it's no problem (bias is cranked way negative).

[Cliff Schecht]

Tried a different transistor (TIP47) with no better results. I also tried varying the amount of current fed into the Zeners with no luck. Still floating at about 520V.. I would love to shed another 100V.. Nobody has any experience with this circuit? Even with a standard CT transformer?

[Ian444]

Looks to me like the 100R resistor and zener need to swap places?

Edit: nope, wrong, disregard...was thinking of regulated power supply.

Scroll down to the bottom of page, same thing almost.

[Super_Reverb]

Tried a different transistor (TIP47) with no better results. I also tried varying the amount of current fed into the Zeners with no luck. Still floating at about 520V.. I would love to shed another 100V.. Nobody has any experience with this circuit? Even with a standard CT transformer?

Cliff, the circuit you are building is a shunt regulator, so...

1) connect the base of the NPN transistor to the anode of the zener diode, one end of the resistor. You are operating the diode in reverse breakdown mode - one way, the diode will drop ~ 0.6V, the other way, it will operate as breakdown diode - positive voltage goes on the cathode.

2) connect the emitter of the NPN transistor and the remaining end of the resistor to ground.

3) connect cathode of zener diode to the collector of NPN. this is the circuit input.

Circuit will regulate collector to emitter voltage to Vzener + Vbe (0.6v). Zener diode current will be set by --> R=0.6/Iz = 0.6V/5mA = 120 Ohms.

This is a simple feedback circuit - feedback from collector to base is through the zener diode. As collector voltage tries to increase, Zener current increases, which forces increase in NPN Vbe voltage, which increases collector current, which decreases collector voltage.


cheers,

rob

[Cliff Schecht]

Finally got it working. It took...count em....ten 24V 0.4W zeners to drop the voltage down to about 450V. I went with R.G.'s circuit using an IRF330 (400V, 5.5A n-ch in TO-3 package) that I found today. I wish I would have tried more zeners sooner but I at least understand the various B+ dropping circuits and their downfalls now. It was only after I put in a 140V worth of high wattage zeners in the centertap and only lost 80V before I realized the zeners were the problem. Lesson learned, now time to make it look pretty.

[xtian]

I'll likely have to drop voltage when I rebuild this Bogen amp. So please sketch out your findings for me. Thanks, Cliff!

[VacuumVoodoo]

Something tells me you tested the orignal circuit without load on the B+ output from the rectifier i.e. STDB switch open in your drawing. This won't work. With nothing drawing current from the rectifier there is no current flowing through the Zener and the transistor, everything is sort of hanging in the air.
I don't understand what you did with all those Zener diodes where you said you achieved some voltage reduction. Parallel connection of Zener diodes is not recommended, they all have slightly different knee voltage so there will always be one carrying majority of current. If you used a series connection then it makes even less sense and supports my suspicion of open STDB switch: your meter completes the circuit by drawing a couple microamperes of current but it's not enough to make the zeners work.

Anyway, here is my verification that the original circuit works when output is loaded, in this case pulling ca 100mA. The two 56 Ohm resistors in series with the diodes are simulating a tube rectifier signifying a hybrid bridge rectifier as originally posted.
Also notice I'm showing you how to derive nice well regulated negative bias voltage directly from the Zener circuit without resorting to additional dedicated bias rectifier circuit.
A note to all patent addicts: You can't patent this. I am hereby claiming priority by placing this in public domain for anyone to use freely, so be my guest.

[Ian444]

Finally got it working. It took...count em....ten 24V 0.4W zeners to drop the voltage down to about 450V. I went with R.G.'s circuit using an IRF330 (400V, 5.5A n-ch in TO-3 package) that I found today. I wish I would have tried more zeners sooner but I at least understand the various B+ dropping circuits and their downfalls now. It was only after I put in a 140V worth of high wattage zeners in the centertap and only lost 80V before I realized the zeners were the problem. Lesson learned, now time to make it look pretty.

??? What's the problem with the zeners?
The whole idea of using zeners with a mosfet or tranny is the active device dissipates the heat, not the zeners. It will need a heatsink.

[Cliff Schecht]

@Alex: It's hard to be clear when talking about such detailed circuits without missing some details. I actually was always testing under load with the standby switch on, I understand that the circuit won't work for shit with the standby off . I think I even mentioned pulling up simulation software and running a full simulation, 5U4's and all, and verified that I was hooking everything up correctly and running everything at the right voltage. In retrospect I think the problem I ran into is that the zener current is not constant. If you look at the waveform at the source of the MOSFET, the current is not continuous but looks more like a rectified sine-wave (as one would expect) which is why the zeners weren't acting as they should. If the bias that was being applied was constant then I think the 5x zeners would have worked fine, but since the current is discontinuous the zeners don't work at their full zener voltage. I'll post a complete schematic below for reference, it's essentially RG's circuit from MOSFET follies.

With the power zeners, I was putting them in series. I know they don't play well together in parallel.

Thanks for the great reply though, it's much appreciated that you took the time to run the circuit sims and help me verify the circuit.

@Ian: The IRF330 mosfet (TO-3 package) is connected straight to the aluminum chassis. It gets hot to the touch after 30 minutes of playing but is not excessively hot. It should be fine even after hours of playing and if it does start heating excessively, I can add a fan and external heatsink to further assist heat removal.

[Cliff Schecht]

I actually think I might have had a bad zener somewhere in the string. I built a new permanent circuit and ended up using 6 of the 10 zeners in the string. This is much more consistent with what my calculations and simulations . Amp sounds sweet, glad this nightmare is over!

[VacuumVoodoo]

If you look at the waveform at the source of the MOSFET, the current is not continuous but looks more like a rectified sine-wave (as one would expect) which is why the zeners weren't acting as they should.

That's normal, the pulsating current is what recharges the smoothing capacitor on the output. It won't affect anything as long as you keep that current loop separate form the rest of your grounding scheme. In other words, don't connect any signal grounds to the left of smoothing cap in your schematic and it will work just fine. Break this rule and you'll hear a wasp buzzing.

[Cliff Schecht]

Yup I was careful about grounding all of the power stuff far from the inputs. The amp is nice and quiet minus whatever crap the carbon comp resistors add. More than usable for the studio and live. Here's some pics of the power scheme and grounding buss bar. Also notice how massive the 5V and 6V heater taps are. The transformer doesn't even get warm to the touch after 30 minutes of hard playing .