I’m about to have a go at building an Aleph 30 clone, courtesy of rhthatcher’s excellent boards and group buy. For the power supply, the PSU board has a nice layout for a CRC filter. So I’ve begun thinking about other parts, not least a suitable transformer, which, as in the ‘standard’ First Watt PSU, should have 18V secondaries or thereabouts, to give rails around 24 - 25 volts.
The thing is, I have several chunky toroids sitting in a cupboard here. So I started wondering whether I could use one (or more) of them. The only problem is there are none with the correct secondary voltage – the lowest I have is 25v.
So, it occurred to me, how about a choke input filter? Maybe with a choke input, one of the transformers I have would yield the right sort of rail voltage. And if I don’t need to buy a transformer, I could spend the money on chokes instead, and make an LCRC filter or even LCLC.
So, I’ve been reading up on choke input filters, and playing with PSUD2. It’s been very educational! There are Hammond chokes that could work, even as little as 10mH, but preferably 20 or 30 mH. There are downsides, though. (Aren’t there always?) One is low frequency ringing. PSUD2 shows that it is quite easily generated, getting more pronounced as the choke gets bigger or the DCR gets lower.
I discovered another problem, though - the ease with which the circuit hits the rectifiers with huge voltage pulses if the load current isn’t quite big enough. According to PSUD2, 900 volts or so can easily be generated if the load isn't sufficient.
A bit of reading on the forum here uncovered some discussion of the issue, and a suggested solution - a suitable film cap (around 0.5uF probably) connected across the input of the choke. Modelling it in PSUD2 (as C1) demonstrates that it ought to work. It also seems to reduce the low frequency ringing.
But the plot thickens at this point. Because at first I entered this cap as 0.5 uF with an ESR of 100 milliohms – a figure plucked out the air. But then I wondered if that was anywhere near correct. Film cap data sheets don’t make it easy, because the ones I’ve been looking at give ESR (or dissipation factor) numbers only for higher frequencies – 1kHz minimum, or maybe 100kHz. And when I guestimate and work out the numbers, they’re much higher than I expected. For example, the Panasonic ECQ-UA caps have among the lowest dissipation factor figures I’ve found, for this size of cap: 0.1% at 1kHz. I’m assuming you can halve that if you’re reducing the frequency by a factor of 10. But even so, when you crunch the numbers, it appears that an 0.56uF cap will have an ESR of 1.42 ohms at 100 Hz.
And that’s where things get really interesting. Because when I put those numbers into PSUD2 and re-click the ‘simulate’ button, something unexpected happens - the rail voltage drops significantly. (Also the low frequency ringing is almost completely gone.)
If PSUD2 is to be believed, a transformer with 35v secondaries would give a rail voltage not of 31v or so (as it would otherwise be), but around 25v. Which is fine for a Aleph 30.
Now, I happen to have two nice 300VA transformers with 35v secondaries. Would using one (or both) be a decent plan? Or is this a bad idea? I’m very aware that there may be things I’m missing or have got badly wrong, and these may be painfully obvious to those with more knowledge/experience than me. If so, let me down gently please!
The thing is, I have several chunky toroids sitting in a cupboard here. So I started wondering whether I could use one (or more) of them. The only problem is there are none with the correct secondary voltage – the lowest I have is 25v.
So, it occurred to me, how about a choke input filter? Maybe with a choke input, one of the transformers I have would yield the right sort of rail voltage. And if I don’t need to buy a transformer, I could spend the money on chokes instead, and make an LCRC filter or even LCLC.
So, I’ve been reading up on choke input filters, and playing with PSUD2. It’s been very educational! There are Hammond chokes that could work, even as little as 10mH, but preferably 20 or 30 mH. There are downsides, though. (Aren’t there always?) One is low frequency ringing. PSUD2 shows that it is quite easily generated, getting more pronounced as the choke gets bigger or the DCR gets lower.
I discovered another problem, though - the ease with which the circuit hits the rectifiers with huge voltage pulses if the load current isn’t quite big enough. According to PSUD2, 900 volts or so can easily be generated if the load isn't sufficient.
A bit of reading on the forum here uncovered some discussion of the issue, and a suggested solution - a suitable film cap (around 0.5uF probably) connected across the input of the choke. Modelling it in PSUD2 (as C1) demonstrates that it ought to work. It also seems to reduce the low frequency ringing.
But the plot thickens at this point. Because at first I entered this cap as 0.5 uF with an ESR of 100 milliohms – a figure plucked out the air. But then I wondered if that was anywhere near correct. Film cap data sheets don’t make it easy, because the ones I’ve been looking at give ESR (or dissipation factor) numbers only for higher frequencies – 1kHz minimum, or maybe 100kHz. And when I guestimate and work out the numbers, they’re much higher than I expected. For example, the Panasonic ECQ-UA caps have among the lowest dissipation factor figures I’ve found, for this size of cap: 0.1% at 1kHz. I’m assuming you can halve that if you’re reducing the frequency by a factor of 10. But even so, when you crunch the numbers, it appears that an 0.56uF cap will have an ESR of 1.42 ohms at 100 Hz.
And that’s where things get really interesting. Because when I put those numbers into PSUD2 and re-click the ‘simulate’ button, something unexpected happens - the rail voltage drops significantly. (Also the low frequency ringing is almost completely gone.)
If PSUD2 is to be believed, a transformer with 35v secondaries would give a rail voltage not of 31v or so (as it would otherwise be), but around 25v. Which is fine for a Aleph 30.
Now, I happen to have two nice 300VA transformers with 35v secondaries. Would using one (or both) be a decent plan? Or is this a bad idea? I’m very aware that there may be things I’m missing or have got badly wrong, and these may be painfully obvious to those with more knowledge/experience than me. If so, let me down gently please!
Last edited:
PSUD is extremely helpful app, proven numerous times
but, as they say - Proof is in Da Pudding ....... meaning - let's see can I make In Vivo what PSUD told me it'll work
so, in most cases, main issue is to get in praxis basic premises you did use in PSUD sim - meaning on power up behavior of load side ( actual amp channels) etc
anyway , main thing for choke input filter is to use caps with big enough voltage, to endure not-still-fully-biased-amp condition, and it really depends of actual amp
say that Aleph circuit is fast in biasing department;
amps having recent opto biasing modules are more than slow in that (M2 and derivatives)
but, as they say - Proof is in Da Pudding ....... meaning - let's see can I make In Vivo what PSUD told me it'll work
so, in most cases, main issue is to get in praxis basic premises you did use in PSUD sim - meaning on power up behavior of load side ( actual amp channels) etc
anyway , main thing for choke input filter is to use caps with big enough voltage, to endure not-still-fully-biased-amp condition, and it really depends of actual amp
say that Aleph circuit is fast in biasing department;
amps having recent opto biasing modules are more than slow in that (M2 and derivatives)
You can easily rplace the R in the CRC with a low DCR (fraction of an Ohm) choke. I'm running CLC on my SissySIT and F5 V3 Turbos I built for a Friend. I used a Hammond
Sissy: 10MH / 5A / 0R16 Inductor Hammond 159ZJ DC Filter Choke. 4 chokes total for a dual rail decoupled power supply.
Turbo: 2.5MH / 10A / 0R044 Inductor Hammond 159ZL DC Filter Choke. 4 chokes total, one per rail x 2 mono blocks.
Bigger question, though, seems more about getting to the right rail voltage using the donut(s) you have on the shelf. With a CLC or CRC you can multiply your secondary voltage by ~1.3 to get the loaded rail voltage.
What is the VA rating of the 25V donut(s)? You could change some values on the Aleph boards and make an Aleph 5 or 60.
The 35V donuts are the right voltage for an Aleph 2, but 300VA isn't big enough. And you need a LOT of heatsink.
Sissy: 10MH / 5A / 0R16 Inductor Hammond 159ZJ DC Filter Choke. 4 chokes total for a dual rail decoupled power supply.
Turbo: 2.5MH / 10A / 0R044 Inductor Hammond 159ZL DC Filter Choke. 4 chokes total, one per rail x 2 mono blocks.
Bigger question, though, seems more about getting to the right rail voltage using the donut(s) you have on the shelf. With a CLC or CRC you can multiply your secondary voltage by ~1.3 to get the loaded rail voltage.
What is the VA rating of the 25V donut(s)? You could change some values on the Aleph boards and make an Aleph 5 or 60.
The 35V donuts are the right voltage for an Aleph 2, but 300VA isn't big enough. And you need a LOT of heatsink.
Yes, CLC with Hammond 159zl appeals, but it wouldn't change the rail voltage I'd get from the transformer.
Conventional view is that choke input would give 0.85-0.9 rather that 1.3-1.4 for cap input. With choke input, I could maybe get 22v from the 25v donuts, then, but they are only 150va, so I'd need to use both. Could I use two identical donuts on a single supply, one on each rail, or is that a no-no? (I'd still need to make sure I didn't fry the rectifiers.)
Or, as I said, I could get 25v from a 35v trannie, given the extra voltage drop that the 'snubber' would result in. Unless the sim is misleading me, or there's something else I'm missing.
I'd been wondering that myself for my planned Aleph 30 build as I have two 200VA toroids, each with two 18VAC secondaries, that would be easy to use and better power-wise than my alternative of a single 300VA toroid. I've been running the 18VAC secondaries in series (but in dual mono configuration) in my F3 clone which has an F1 style PSU (0V / 50VDC). The Alephs as you know require an "F5 style" PSU to deliver +/- 25VDC.
The schematic below shows an example of a +/- 60V PSU that uses two toroids. The two trafos look to me to be in series on the DC side of their bridge rectifiers, with the midpoint of that series connection serving as the 0V / ground pole for DC. What bothers me is that tying the secondaries of two trafos together in AC space would seem to require them to be perfectly in phase? I don't know if tying them together in DC after rectification as is shown in this schematic gets around this or not, and I'm a little too chicken to try.
Attachments
what I wrote above is to wire them exactly as shown in that schm
then you don't have any problem
then you don't have any problem
Well, I've talked myself out of this, for the time beng at least. When I started thinking about it, it was hoping it wouldn't cost much extra, as I'd save the cost of buying a xformer. But when I do the sums, it doesn't really add up the way I hoped! I'll go with the usual CRC for the moment, and maybe try chokes in the future, depending on funds.
Randy, I'm interested that you've tried both the 159ZJ and the 159ZL in CLC applications. It's an interesting choice between higher inductance (the ZJ), and lower series resistance (the ZL). Both model pretty well, but did you form any impression about which is preferable?