Aleph-X PS question

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
So I wanna build myself a frugal Aleph-X. The main area to cut cost is the use of surplus EI transformers (2 x 12VAC) and surplus caps for the PS.

The thing that scares me is how do I tweak out any PS caused hum if it comes to that? As I understand it, a pi filter is not easy to achieve because a very low dcr inductor is needed due to the high current draws involved. So, what kind of tweaking can I fall back on if hum turns out to be an issue?
 
Leadbelly,
I am doing exactly as you describe- cheap surplus trannies and caps.
I will warn you now...
count on your transformer voltage being lower than advertised, and your dc voltage- ac voltage ratio being right around 1:1.
I bought 4 13.5V@20A transformers. Unfortunately, I found out that the 13.5V is the unloaded value. Depending on regulation, I am hoping to get 12-13V out of them, DC. Ideally, it would be more like 15-16V. I will eventually replace these with some higher voltage toroids.
A lot of the noise will depend on how much current you are pulling out of the transformers. I will have a 7A constant current out of each of mine. I have found that the only way to really get the ripple down there is to use a pi filter. As you said, you will lose a lot of voltage this way. I plan to use 2mH 14ga. air core inductors. These should be about .31ohms for a drop of a little more than 2 volts.
If you haven't done it yet, download the psu designer from duncanamps.com. Play with it for a while, and see how different factors (transformer regulation, capacitor ESR, series resistance, etc.) affect the outcome. This will give you a little more of an idea of what you want to do. Hope this helps.
-NS
 
Reducing PS ripple

I'm currently assembling my Pass Labs a-40 amp and have noticed some hum at the speaker that is the result of power supply ripple. My original configuration was toroid -> bridge -> caps (24,000uF per rail, surplus caps). When I placed a CRC pi-filter in the power supply, the hum was gone, and the speakers were absolutely silent!

The new power supply configuration is toroid -> bridge -> 24,000uF cap -> 0.30 ohm 5w wirewound resistor -> 24,000uF cap. The resistor gets a little warm, but certainly not hot and it drops less than one volt for each rail.
 
The new power supply configuration is toroid -> bridge -> 24,000uF cap -> 0.30 ohm 5w wirewound resistor -> 24,000uF cap. The resistor gets a little warm, but certainly not hot and it drops less than one volt for each rail.

CRC's can be effective, but a choke is more effective, given the same series resistance. I looked into doing a similar thing, but found the difference in ripple would be about 80mV (or more) with a CRC compared to 8mV (or less) with an inductor of a similar resistance. That's 10X better!
My supplies will consist of 12V surplus tranformers, bridge pack, 10 X 6800µF caps, 2mH inductor (.31 ohm), 10 X 6800µF caps. Should be about 5mV ripple.
 
Reducing ripple

True, using a choke may reduce measureable ripple to a greater extent than a simple resistor in series with the caps. However, once the hum is gone from my speakers (as it now is with the resistors), I don't really care what measured value it has:D It is possible that speakers with greater efficiency may require something more than a simple resistor, though - I'm clearly speculating now.

Additionally, I can buy a resistor for each of my rails and not spend more than $1. Inductors that can handle the currents needed in a power supply are :$: :$:
 
You're scaring me off here. To my thinking, having to buy big inductors to make a quiet PS will negate the cost savings of a surplus transformer in the first place.

Is there something else I could look at, like maybe a higher order filter using just R's and C's? Or a capacitance multiplier circuit?
 
leadbelly: Looks like your jumping over board before you know for sure that the ship is sinking :eek:

Here is what I would suggest. Gather all of your parts and hook them together on the table top and give it a listen (don't assemble the parts into a custom chassis just yet). It may turn out that you don't need anything other than a few caps to get an amp you are happy with. If that's the case, then you are done! If, on the other hand, you have some noise that seems to be coming from power supply ripple, try the cheap solution first: put a resistor in series with a second cap to make a pi filter. Chances are, this will remove any audible ripple induced hum that you have. If not, then get your inductors (after making sure that your hum is not coming from some other source, such as induced by RF emissions from your power supply)

This is why its important to hook it all up for the first time with the parts spread across the table top, not crammed into a chassis. If you have hum, you can play with moving wires around, moving transformers closer/further away to see if that cures your problem. Then, when you are happy with the sound, fabriate your chassis. This way you don't end up in the situation of not having room to add extra components, should they be required.
 
You can also roll your own inductors using a home made form. I made a form out of a 1" broom stick handle (the core) and two thin slabs of 4" square wood to make the sides. This was all bolted together with a long bolt that goes into the chuck of a variable speed electric drill. I takes about 150" of #14(I think) of enamaled wire wound onto this form to get around 2mh with DCR of .5 ohms. Once wound, unbolt the bolt, the sides come off, and punch out the wooden core. I used nylon ties to hold the coils. Cost of wire is between $10 and $15. per coil. All the hum went away when I put the coils between the capacitors in my aleph 3's. Well worth the effort.

I like to think that the inductors also keep out the high frequency hash. I also like to think that the coils add a bit of warmth to the amps.
 
nobody special said:
Leadbelly,
I am doing exactly as you describe- cheap surplus trannies and caps.
I will warn you now...
count on your transformer voltage being lower than advertised, and your dc voltage- ac voltage ratio being right around 1:1.
I bought 4 13.5V@20A transformers. Unfortunately, I found out that the 13.5V is the unloaded value. Depending on regulation, I am hoping to get 12-13V out of them, DC. Ideally, it would be more like 15-16V. I will eventually replace these with some higher voltage toroids.
A lot of the noise will depend on how much current you are pulling out of the transformers. I will have a 7A constant current out of each of mine. I have found that the only way to really get the ripple down there is to use a pi filter. As you said, you will lose a lot of voltage this way. I plan to use 2mH 14ga. air core inductors. These should be about .31ohms for a drop of a little more than 2 volts.
If you haven't done it yet, download the psu designer from duncanamps.com. Play with it for a while, and see how different factors (transformer regulation, capacitor ESR, series resistance, etc.) affect the outcome. This will give you a little more of an idea of what you want to do. Hope this helps.
-NS

My practical experience:
two cl-30 in parallel on transformer primary.
600VA toroidal from VMagnetics, dual 12.5 VAC secondaries, dual bridge, followed by 230,000uF+ 2x0.33 ( in parallel), another 230,000, the Pi filter part is obviously doubled for the negative side.
The caps are joined by a Cu plate about 1/32 inch thick 4x4 inches long and wide to form the ground. Two leads from the bridges go to GND at the center of this plate, the grounds from rev.1.0 board go to this center GND as well via 16 gauge OFC wire exactly 5 in long each.
Results:
-the PS voltage with 10 AMP draw drops to 11.75VDC which is a little lower than I like. Remember I have a 0.165 ohm in the pi filter.
-there is no hum to speak of, no GND loops.
-no turn on/off noises
-sound: amazing

As an aside, I am driving the AX balanced with a BOSOZ the AX is DC coupled with 10 kohm input resistors. So I have only the coupling caps of the BOSOZ and I am going to change the 10k resistors to ground as well or maybe get rid of them altogheter.
Also I have no over current protection installed on the output gain mosfets.
The DC offset stays put while playing music.

All things considered this is a very stable design.
Keep in mind that it draws alot of current and that transformers settle A LOT especially crappy transformers.
IMHO inductors are a waste of time and money here. At 0.3 ohm they end up dissipating too much power and drop the voltage too much.
 
My vote would go to a capacitance multiplier. I've had good results using it for a Class-A headphone amp (the Szekeres).

Leadbelly, I see you're from here. I have a source for surplus caps; if you're interested, send me an email.
I've ordered some surplus transformers from Apex Jr. that I'm hoping to someday use for an A-X. It involves putting the primaries in series, so there may be some surprises. We could collaborate on a table-top experiment if you want.
 
chris ma said:
grataku,

are you saying your pi-filter is C-R-C and your R is the 0.165ohm?
Chris

Now I get it I think. 2X0.33 in parallel, what watt are these resistors are rated?

chris ma
25 W each. I am dissipating about 32W total for + and -.



Paul
you can do as you please but IMO if you are into S&M there is a much more fun way to do it (so they say). ;)
Capacitance multiplier would be very painful to implement here, in order to work reliably you are going to have to have it's own dedicated heatsink which is going to have to be as large as the amp heatsinks.
 
grataku said:

Paul
you can do as you please but IMO if you are into S&M there is a much more fun way to do it (so they say). ;)
Capacitance multiplier would be very painful to implement here, in order to work reliably you are going to have to have it's own dedicated heatsink which is going to have to be as large as the amp heatsinks.
Dedicated heatsink, agreed, but I doubt it would have to be as large as the amp's. I figure you'd drop a couple of volts or so, at 4 amps that's maybe 10 watts per pass transistor.
I'm planning for a separate power supply enclosure.
 
leadbelly said:
Thanks to all.

I think my 1st rev will be 12.6 VAC xfrmr + 2 x 10,000uF + 2 x 0R33 + 3 x 10,000uF + 10uF bypass and I'll see how that works.

Also, what kind of cheapo connectors can I use between my PS case and my amp case? I'm not sure what to use.


I would put more caps before the 0.33 ohms which is where the bulk of the ripple current goes through, so helps to have the the lowest ESR there. I would also start from 14VAC instead of 12.6.
I suspect it will be a little more noisy than if using 230,000 caps ;). You should try to get your hands on bigger caps if you can or you can drop more voltage across the resistor and start from a 16VAC trafo, up to you.

You may want to thing about NOT going cheapo here or NOT to go with separate enclosures at all, there are lots of amps going around I would avoid long wires and connectors if at all possible.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.