[AtomBoy]

I'm planning a low-buck P2P/Dead Bug scrappy Borbely Hybrid.

The article is here:

http://www.borbelyaudio.com/pics/405borbely2506.pdf

It calls for dual power supplys for left/right of +-24V 100mA or +-15V 160mA and a third +6.3V 350mA (I'm planning on 700mA for both) for the tube heaters.

You can buy the recommended kits from Borbely which I'm sure are fantastic, but I'm going cheap-o caveman here. Between looking at the data sheet for the LM337, Tangent's STEPS, and some Google hits I've cooked up the following:

T1 = TE62034-ND TRANSFRMR TOROID 18V .388A
T2 = TE62031-ND TRANSFRMR TOROID 9V .778A

Most of the rest of the part names and values are from Tangent's STEPS parts sheet:

http://www.tangentsoft.net/elec/teps/plists.html

I've left some parts out for the sake of simplicity and budget.

Questions:

1) What am I buying with the dual PS? Is it really worth it if they're sharing the rectifier and the filtering caps? It's not a hell of a lot of parts, so I'm inclined to just do it.

2) Where is the sweet spot on the filtering caps? I'm going to start with 4x1000uF 50V and leave room for more.

3) Have I left out a $.50 part that would make a huge differance?

4) Is the TO-220 the right package for the LM137s and LM337s?

5) What bone-headded error did I make in transformer selection/ layout/ other that will result in tears/ insurance claim/ obvious hum?

Cheers!

Edit: The diodes are pointing the wrong direction on both rectifiers.

 

[Tomo]

Hey,

1) I am pretty sure if you are only interested in the cost, the price will be eventually be the same. ... But if you still wanna do it, go do it. ... I think we confuse you too much.

2) Filtering caps? You talking about pre-regulator or post-regulator? For pre-regulator stage, 2200uF-4700uF is standard. You may increase but make sure your transformer and rectifiers are strong. Some people do 10,000uF, but it may not be cost-effective. For post-regulator stage, it all depends on the amp type. Standard is 100-500uF right at the output of PSU. Generally I put 100 or 220uF here because this value is very common and easy to get good deal on them. You may want another 100-500uF on the amplifier rail. Your amp has push-pull stage which might like more power available.

3) You are not missing parts I think.

4) LM317 and LM337 are available with TO-220 casing. TO-220 is probably easiest to craft with.

5) You are no bone-head. ... None of us are

(I am not talking about Diode Directions ... you figure that one out.)

Looks pretty spiffy to me. But I would simply a lot since I am a lazy bum. I am so lazy, I picked up those PSU kits and spiffed them instead. ... I am also a DIY-Heretic, as you can see.

Tomo

P.S. You might think my post-regulator capacitance is a bit low. ... But I realized that placing and quality of caps here is a lot more important than the quality. ... I got away with 10uF once.

P.P.S. You might want to install AC-Line-Filter. Corcom Filter is a good example, but you can use other stuff. They are fairly cheap and good habit to install. I am using bulky metal can package, but socket-type filter is extremely easy to install and will look very professional.

 

[rickcr42]

You do not need a dual mono supply to power this amp,or any amp.It is more taking it to the next level and if expense is a factor the returnb vs. cost not really there.
think of it more as a tweak and taking the amp to the next level (as would be true of any stereo device) but not as an essential for operation or good sound.

I would build a single supply then if later you want to ramp up to two it would be a simple addition.Why complicate the simple right out of the gate ?

 

[AtomBoy]

Quote:

Originally Posted by rickcr42 You do not need a dual mono supply ...

I didn't think I had to have the dual supply, but I was wondering what it was supposed to do. Better detail?

I was also wondering if not having a seperate xformer-rectifier-filter cap section was defeating the purpose of two supplies.

I'm still doing some reading on transformer selection, but as of right now I think I need a little more amperage.

 

[erix]

Wow - this headphone amp looks almost EXACTLY like this linestage!

Why don't you just buy this one?

ok,
erix

 

[rickcr42]

Quote:

I didn't think I had to have the dual supply, but I was wondering what it was supposed to do. Better detail?

I have a raging party at my house right now so I hope this response makes some sense :

With loudspeakers where there is a natuaral left/right cross and it is a nice touch to have totally separate amps for each channel where having each amp have its own power supply means the image is locked down and not dependant on which channel is dominating the current draw from the supply at the moment leading to a fuzzy image.

With headphone amp the image is all in your head and since the open air natural crossfeed of the loudspeaker is not present any interchannel blending is a plus so a single power supply with say double the overall current requirements for the entire amp is not only good enough but in this case,headphone use a plus.

By overbuilding the PSU dynamics will not suffer or the image wander and any help the stereo signal meant for loudspeakers and following the blumlein model which takes into consideration both microphone crossfeed at the recording end and loudspeaker blending at the monitoring end is not a bad thing/total isolation and total power supply reserve for each channel is meaningful in the realm of loudspeaker driving and at the front end (RIAA stage,Line stage) but not so much at the end stage,the headphone driving stage if you allow for current reserves to not only power the stage but power both channels during a musical peak event.

Hope this helps and is semi lucid

Rickamudo

 

[rickcr42]

Quote:

Wow - this headphone amp looks almost EXACTLY like

might have something to do with it being the same amp

A good line stage does not always translate well as being a good headphone amp but a good headphone amp many times makes a great line stage.The requirements are mostly identical except for the ability to drive a low impedance which is where most line stages fall short.

32 ohms is no easy load and even 600 ohms a real bit*h when the building for a 10K or better load which puts little strss on the output stage of the line amp

 

[AtomBoy]

Quote:

Originally Posted by erix Wow - this headphone amp looks almost EXACTLY like this linestage! Why don't you just buy this one? ok, erix

Thanks for the link, Erix, I'll keep an eye on it. I'll be interested to see what they use for parts. I'm not doing this so much for the end product, as for the journey and nothing makes you look at a schematic like doing point to point. I'm learning a lot on every step of this project.

Quote:

Originally Posted by rickr42 I have a raging party at my house right now so I hope this response makes some sense :

I'm recovering from a party, so I'll try and make sense of it later!

For right now, you've talked me into it (single regulator stage for +-15V). I'm still looking into right-sizing the transformers.

 

[AtomBoy]

Hey, it worked! I went with a single +-15V side.

It might not sound the cleanest- I won't know until I build the amp, but it should be good enough for a start. This dead-bug stuff has more in common with jewelry making than anything else. Anybody have any advice for dead-bugging?

The top is a 3/16 Al plate, the sides are steel from an ancient Steelcase desk that was on its way to the dump, lots of bits are from Surplus City including the big filter caps, the terminal strips and the pots. There are three filter caps- one is hiding under another in the picture.

Now to build the amp...

 

[AtomBoy]

I've built one channel:

I have only had the courage to use the crappy tubes, plug it into the crappy source and the crappy earbuds and of course it sounds like crap. It makes more sound than smoke so I'm not too discouraged.

I can't get the DC offset below 0.7V after its warmed up. Before the heaters are warmed up- look out! I've fried one set of cheap-o earbuds so far. They were too hot to touch. Neat! I guess I'll check all the resistor values again, but I'm pretty sure they are right.

I did run it for a while with the pins swapped on the P-Channel MOSFET. It's mirrored in the schematic and I numbered the pins as though it weren't- d'oh! It made no sound and one of the little transistors got pretty hot.

I may just build the other channel and see if it does the same thing. If it does, I have found a few other variations on this circuit (TubeCAD.com, etc. - the links are on another computer- I'll edit them into this post later) that might be worth a try.

I would certianly never put this thing into regular use without a protection circuit. I'm looking at AMBs Epsilon12. ( This Thread. )

Other variations on this theme:

http://digilander.libero.it/essentia...id_circuit.htm

http://www.tubecad.com/2004/blog0001.htm

 

[erix]

EXCELLENT DEAD BUGGAGE!

impressed,
erix

 

[AtomBoy]

Well, I put the other side together. I'll admit I'm in over my head, but I'm learning something.

This time I can adjust the DC offset to zero after giving the tube about 5 min to heat up and it seems to sound less distorted than the first try. It's still not at all hi-fi. It sounds like something is being driven too hard. There's also a buzz that comes and goes.

When I was running both channels, it sounded much better. I'm wondering if its because the ps was delivering all the current it could and the MOSFETs weren't being overloaded?

When I measure the current its pulling, I get .156A on both the + and - side and .003 on the ground for just the one channel. This is what the article says I should see. I cut the power leads to the first channel to see if the second would get as hot It doesn't get *as* hot, but its still pretty toasty.

With both channels powered, it draws about .36A and the 6 x 9 in 3/16 al plate gets almost too hot to hold on to. Judging from the puny heat sinks that the comercial kit versions of this thing come with- I'm making way too much heat.

I am seeing some weird stuff in the power supply. My 36V @.35A transformer is giving +25 and -25V right before the regulator. The regulators get pretty hot as they drop 10V. The transformer warms up pretty good as well.

I know it looks like a dog's breakfast, but its actually pretty solid as long as you don't poke anything. It took much less time to do this second side- about 2 hours.

The only thing I can think to do now is to bump up the value for R13 that sets the current used by the MOSFET. If that doesn't cool things down a lot, I may just declare victory and retreat to another variation on this theme.

Are the voltages coming out of the transformer normal?

 

[Voodoochile]

Although it's within spec, that's a lot of voltage to drop across those regs. They are cooking right along. Also, you are apparently drawing .36A through a transformer rated for .388A Don't forget the voltage gain from going through a full-wave bridge is going to get you right up there. A dual 15v should be fine with a full-wave bridge to deliver +/- 15v. Vout x 1.38 (FW bridge) = ~voltage presented to the regulator. With your +/- 18v tranny, you are getting about 24.84v, and having your sweaty regulator whittle it down to 15v. At least you have a huge heatsink!

Perhaps the 15va TE62053 would be a better choice than the 7va unit you have. It could handle the current better, and would be dropping less voltage across your regs. 15v x 1.38 = 20.7vdc. Probably abit more, as the 15va toroid will be less loaded down. You do want it fairly well loaded, but when you start up and charge your caps, that toroid you have is working overtime.

Nice p-t-p project!

 

[AtomBoy]

Quote:

Originally Posted by Voodoochile Don't forget the voltage gain from going through a full-wave bridge is going to get you right up there.

Thanks, Voodoochile. Maybe that's where I'm getting bitten. I didn't realize there was a voltage gain with the rectifier. I was expecting the regulators to see +-18V minus a voltage drop from the diodes. If anything, I was worried about not having enough voltage after drops from the diodes and regulator but then I read something about transformers delivering more than rated voltage when they're not loaded. It's the VA rating that you really get so pulling less A means more V?

I've been stumbling through 'The Art of Electronics' but didn't get much enlightenment. Can anyone point me at some good reading?

Edit:

The following seems on target:
http://www.mcitransformer.com/i_notes.html

This has a little more background:
http://www.st.com/stonline/books/pdf/docs/1707.pdf

 

[Voodoochile]

Here is the full bridge rectifier formula, as I know it now:

1. Convert Vrms to Vpk by multiplying by 1.414 = 31.1V

2. Subtract *two* diode drops because in a bridge rectifier there are two diodes in series during any conduction period, and assume 0.7V per diode (Vf varies from 0.4 to 1.1V depending on diode type and PIV rating)

This is also very worthwhile reading:
Quote:

There are equations to determine the amount of capacitance, and there are rules of thumb. People around here tend to use an exorbitant amount of capacitance to achieve the lowest ripple possiblle, but this is emphatically not the best way to go because it greatly increases the peak to average current ratio flowing through the circuit. So what? Well, high peak currents can cause all of the following bad things: 1) increased radiated magnetic noise; 2) increased voltage spiking from stray inductance; 3) reduced diode and capacitor life; 4) higher ripple, even, both from the magnitude of current and the self-heating it induces in the capacitor (which raises it's ESR). So if more is not the merrier when it comes to capacitance, and too little is obviously not either, then what is the happy medium? Well the first rule of thumb for regulator circuits is simply to make sure that the trough of the ripple voltage exceeds the regulated output voltage plus the regulator's drop-out voltage. Thus, the higher the unregulated supply voltage is compared to the regulated output voltage, the less capacitance will be needed for a given load current. The flip side to this is that the more voltage the regulator has to drop, the hotter it will get in operation. The rule of thumb is to aim for 5-10% ripple at the input to the regulator and then adjust that value up if the transformer voltage is much higher than the desired regulated voltage or adjust it down if vice versa and/or very low ripple feedthrough is needed (keeping in mind even crappy three-terminal regulators will get you 40dB of ripple rejection, with 60-80dB being more common). So far this is a lot of blah-blah-blah. If you've made it to this point without falling asleep you deserve a worked example. Let's design for 24V at 0.8A average with a 24VAC at 1A transformer. The output from a bridge rectifer will be a peak DC voltage of 1.414 * 24 - ~2V (for the diode drops), or ~32V. 10% of this is 3.2V, and half a cycle with 60Hz mains is 8.33mS. Plug these numbers into the capacitor equation: C = (dt * I) / dV C = (8.33mS * 0.8A) / 3.2Vpp C = 2,083uF Use 2x 1000uF or 4x 470uF - it's okay to go a little below since we aren't even close to the regulator's dropout voltage! It is instructive to see how much capacitance would be required to feed a Class A amplifier with approximately the same power requirements but with such vastly lower ripple tolerance that enough capacitance to knock ripple down to 0.1% is necessary. The relevant numbers become: C = (8.33mS * 0.8A) / 0.032Vpp C = 208,250uF !!!

From the mother of power supply critiques thread.