REDD47 Psu

[Deepdark]

Hi there

I'm about to build a Redd 47 and was looking at the psu. I first found the JBB design with a 6X4 and the 2* 0A2 regulators. Really simple and nice. I redraw the psu schematic with diode rectifier and I replace the 2x 0A2 with a zener string. I would like to know if I'm ok, or if I should put another filtering stage. By the way, I was searching how to evaluate the ripple factor within a solid state rectifier but found nothing really clear about that. I know how to determine it in a tube data sheet with the curves, but i'm not sure with the diode. Thanks

 

[Deepdark]

hmm looks like the schematic don't work. Here is linked the JBB schematic

 

[Deepdark]

oh and by the way, if anyone can tell me a little about the 0A2, are they good as regulator? How do they work?

 

[My3gger]

Hi, my experiences with voltage reference tubes like OA2 aren't too good because of noise. IIRC an early Drip's version of Redd 47 had mistake in this part, watch for that too. There is a lot of reading online, like this fom Steve Bench:
http://www.jacmusic.com/techcorner/SBENCH-PAGES/sbench/reg2.html
Purple glow is nice, but why bother in new builds if zeners are cheaper and effective.

 

[Deepdark]

thanks My3gger

Looking at my schematic, do I put the zener strings right? The 470K in paralelle is a load resistor?

 

[PRR]

There are things I do not understand in this plan.

* No first filter cap? (It is traditional; also it makes the initial DC voltage easy to guess.)

* I would ball-park (rough) 240r and 47u for 18.5dB reduction of 120Hz, not 8.5dB. This is not exact though.

* If you have two 240 Ohm resistors, 350V on one end 300V on the other end, isn't that 0.104 Amps? Seems rather a lot for a 0.020A preamp.

* If you do have 100mA flowing in the resistors, and 20mA going ot to preamp, that leaves 80mA going into Zeners. Are these Zeners good for 4 Watts each?

* What if the wall-outlet rises or falls 10%? 20%? (I have +0%/-15% variation here constantly.) This type R-Z network, the dissipation rises much faster than the line voltage, and spare current falls much faster than line voltage.

* (pet peeve) Why regulate?? The tubes are not fussy. You need to filter the heck out of the 120Hz ripple, and enough long-term smoothing so that wall-voltage changes come out sub-sub-audio and are lost in the audio coupling networks. Caps are cheaper by the 10-pack. 350V-300V= 50V to lose. 50V/20mA= 2,500 Ohms to drop the voltage. Take five 500r (470r) resistors. Total diss is 50V*0.020A= 1W, so 1/2W each is ample. Dart your first cap as 1uFd/mA, round-up generously: 40(47)uFd. Take the other caps the same. This gives 5 sections of 8Hz high-cut. Loss at 8Hz will be 5*3dB= 15dB, rolling-off 100dB/decade above, so over 100dB at 120Hz. Even at 4Hz there's 5dB of smoothing so any wall-voltage jumps will be sub-audio. And no breakdown hiss.

 

[Deepdark]

So a typical RC network without the tube/zener regulator. That was my first thought but somewhere over here i read lot of folks claiming that the "particular sound of the redd47" came from the regulated psu...which i quiet dont understand since the goal is to fed the tubes with the proper voltage at a given current.

By the way, what about the référence schematic i linked? The one from JBB? He should encounter the same trouble looking at the current?

Where do you take your 8hz by the way? Whitin the formula 2 x pi x 120hz x 500R x 47UF i got 18.71v convertie in dB = around 25db. So five stage Will give about 125db of ripple reduction. Is that it?

 

[My3gger]

Yes, RC filter. PRR's calculation show it will be very quiet, probably cheapest and easy to build/experiment too. I can't say for sure if regulators really chance the sound, you could try it yourself with simple Gyraf G9 HT psu vs RC filters. There are application where regulation is needed, like vari-mu compressors etc.
JBB's schematic would have 69mA through resistors, check OA2's datasheet to see if this is still too much.
8Hz are taken because ripple is at 120Hz with this kind of rectifier configuration, so it is much lower than lowest "audio" frequency and remaining small ripple won't affect it. Seems like mentioned 5x500r resistors will kill any ripple, voltage reference or real regulator is not really needed for preamp to function properly. Some of us are just playing with other configurations even when RC filter works ok.

p.s.: Check Ian Bell's (Ruffrecords) advice for simple setting of G9's TL783 if you want comparison, this way it doesn't require load when tested and 275V is enough. It is very simple to recalculate other voltages if you check datasheet:

"TL783C data says that it needs at least 10mA load or so to start regulating, so it might be normal that at startup, when cold tubes don't draw any current yet, B+ rises (unregulated) . Emrr suggestion to add a power bleeder resistance for minimum load should solve this.
regards,
Laurent."
No need for a power bleed resistor. I have already done some experiments by modifying my TL783 based phantom power PCB to become a regulated HT board. If you use a 150R from the output to the ref pin and a 33K 5W resistor from the ref pin to ground then you get the required minimum current and a 275V output.

 

[mitsos]

I used a mosfet regulation circuit posted by Kingston in his M670 compressor thread.  I only built one stage, and it worked very well.  The schematic is in the first post of the thread., everything up to and including R6.

http://groupdiy.com/index.php?topic=51478.msg654153#msg654153

Here's a picture of the PCB. The HV is the top half of the PCB, the bottom is a voltage multiplier for phantom power. Heater is done CRC on a terminal strip.

http://groupdiy.com/index.php?topic=51478.msg751793#msg751793

I was seriously blown away by the sound, how much was the redd47 and how much would have changed if the PSU were unregulated, I don't know, but I'm curious enough to test it out some day.

 

[My3gger]

I helped a few people with their Redds where power transformer didn't have enough current which caused several problems, some directly with the sound. Here is Hammond's transformer selection guide.

 

[My3gger]

Ian't pdf about the same thing has some more calculation and is related to Hammond's, might help understanding it better.

 

[ruffrecords]

One thing to remember about class A amplifiers is that their power consumption is by definition constant, with or without signal. When signal is present, on one half cycle the current draw is increased and on the other half cycle it is reduced by the same amount. The average current draw is therefore constant. If the amplifier has adequate local decoupling, the increased current is provided by the local decoupling capacitor which recharges from the HT supply. If the decoupling is adequate then the current drawn from the HT supply is nearly constant at all signal levels. A stabilised supply is therefore has little if anything to stabilise against and in most cases is quite unnecessary.

Cheers

Ian

 

[PRR]

> class A amplifiers is that their power consumption is by definition constant, with or without signal.

Quibble. Definition is usually "no cut-off at any part of the signal swing". Makes no assertion about supply current.

You may quibble my quibble. These "classes" are more text-book than in-the-trench engineering distinctions. The several names and the conditions they were applied to varied over the years and over the ocean.

It is possible to design a Class A (no cutoff) amp where the current does change with signal. In some amps it does. A heavily-loaded 6L6 stage will show a small current rise with signal. There was a car-radio tube where design current *dropped* from high at zero signal to low at full output. Both conditions strongly imply significant 2nd harmonic distortion, and are usually avoided in "good" audio. So *usually* idle current IS constant in "good" audio amps worked class A.

There are also wacko schemes to bias a class A stage cool for idle and hot when signal happens. May not handle initial transients well.

 

[mitsos]

One thing to remember about class A amplifiers is that their power consumption is by definition constant,
Ian

That was my argument to the people who thought they needed a PSU with 1-2A per couple of channels of Neve preamps (EZ1290), when Neve spec was about 120mA per channel, with EQ, which most people didn't build in. In the end, everyone wants a bigger supply. 

To me the biggest benefit of Kingston's circuit is not having to fiddle with voltage dropping resistors, no trial and error, it just works.

cheers! 

 

[Brian Roth]

I always wondered why the EMI engineers added that pair of 0A2 bottles to the REDD 47 modules.  But, if it was Good Enough for the Beatles.....  <g>

Bri

 

[ruffrecords]

> class A amplifiers is that their power consumption is by definition constant, with or without signal.

Quibble. Definition is usually "no cut-off at any part of the signal swing". Makes no assertion about supply current.

You may quibble my quibble.

You are quite right. Many tube based singled ended class A amplifiers have an asymmetric transfer function which is what causes the characteristic second harmonic distortion. This means that the peak current drawn in one half cycle is not exactly balanced by the peak current in the other half cycle and it is this that causes a signal level dependent shift in operating point and a slight change in average current draw.

In typical class A tube audio pre-amplifiers the effect is negligible. Perhaps I should have made the distinction with the typical class B output stages found in op amps where the supply current is very much signal dependent and a regulated supply is pretty much essential.

Cheers

Ian