Voltage should not execeed 25V and I would suggest a double secondaries instead of a center tapped one.
Thanks ClaveFremen. What should be it's wattage? Does 12 amp suffice for a stereo? What's the benefit of a dual secondary as against a center tapped?
You're welcome.
300VA is a good rating for a stereo My_Ref.
That you can reuse it for other projects which have double diode bridge, like my variation, the My_Ref Fremen Edition.
Thanks again.
BTW I notice that this circuit has no output decoupling caps. I have some DIY 3-way speakers and they have a electrolytic cap in series as a make-shift X-over. Should I just remove them all?
BTW I notice that this circuit has no output decoupling caps. I have some DIY 3-way speakers and they have a electrolytic cap in series as a make-shift X-over. Should I just remove them all?
Seems to me that rectifying section sould be separate for positive and negative rails if double secondaries are used such that the secondaries are never connected together. Can't remember what the original MyRef is.
From double secondaries you can easily obtain a center tapped connection as needed by the original MyRef.
It is true, but if the phase of two secondaries are not perfectly the same, you will,hear hum when you do that. At least this happened to me in one of my preamp designs.
Russ had mentioned using an LM4562 in place of the LM318, has anyone tried that? Mother phase looks awfully nice, just the gain is higher and the slew rate is lower.
The single opamp replacement for the LM318 is the LME49710, but it won't work stably in Rev C, E and FE without major changes to the compensation. It may work in a Rev A, but the audible sonics will not be comparable to a Rev C with an LM318 - YMMV.
I've tried the LME49710 in a stock Rev C, and it triggered Spike and other instability immediately.
I am wondering that the compensation may not be necessary for the LM4562, I cannot find the open loop gain/phase data for the LME49710.
I have the power supply now. But I'm still waiting for an answer to my previous question. Does this amp need an output capacitor between the output and the load? Thanks.
Nope, it's directly coupled to the load. However, most builders use an input coupling cap to block the DC at the input.
Tried the classic lines pcb I bought over eBay. Couldn't get it to sound anything. I'm not asking for help with something that I bought over ebay but just a nod whether my testing style was ok.
Conditions:
1. Power: 28v-0v-28v, <= 10Amp
2. Speaker: 4Ohms. I also connected a 3.9Ohms/5w resistor in series with the LS to see if changing the load would have any effect.
3. The audio input was from my android phone at minimum to max level and got not output.
4. The heatsink per LM3886 was around 20 sq cm. aluminum.
Results:
A. The LM3886 got very hot pretty fast. I turned the amp on for ~30sec and I couldn't put my finger on it.
B. Increasing the volume of the input audio didn't produce any output.
C. Trying a audio test by touching the open inputs and it didn't output any audio.
D. I tried to see if I could hear any relay click in or even any pop noise in the speaker when turning on. No audible sound except for the one that the on/off switch made.
Kindly provide any info on how you tested your amps, if I'm doing anything wrong or just point me to any other thread. { I've spent days and nights without food trying to troubleshoot some of my very first projects (SS radio, LA3161 preamp, LA4440 amp, TBA810 Amp) and I remember how painful it was. I'm already getting scared of trying to imagine the number of hours I need to troubleshoot this thing. }
Arranging a chassis for this thing is another long story though. Suffice it to say that the chassis is a 50yr. old new corroded piece used for valve circuits that I had to get painted.
Conditions:
1. Power: 28v-0v-28v, <= 10Amp
2. Speaker: 4Ohms. I also connected a 3.9Ohms/5w resistor in series with the LS to see if changing the load would have any effect.
3. The audio input was from my android phone at minimum to max level and got not output.
4. The heatsink per LM3886 was around 20 sq cm. aluminum.
Results:
A. The LM3886 got very hot pretty fast. I turned the amp on for ~30sec and I couldn't put my finger on it.
B. Increasing the volume of the input audio didn't produce any output.
C. Trying a audio test by touching the open inputs and it didn't output any audio.
D. I tried to see if I could hear any relay click in or even any pop noise in the speaker when turning on. No audible sound except for the one that the on/off switch made.
Kindly provide any info on how you tested your amps, if I'm doing anything wrong or just point me to any other thread. { I've spent days and nights without food trying to troubleshoot some of my very first projects (SS radio, LA3161 preamp, LA4440 amp, TBA810 Amp) and I remember how painful it was. I'm already getting scared of trying to imagine the number of hours I need to troubleshoot this thing. }
Arranging a chassis for this thing is another long story though. Suffice it to say that the chassis is a 50yr. old new corroded piece used for valve circuits that I had to get painted.
1) 28-0-28 is way over spec for the LM3886 - it would be about +/- 38V for the rails, which is pretty close to the absolute max +/- Vs specs for the LM3886. You need a 24-0-24 trafo, or the LM3886 will go into SPiKe protection pretty quickly - which is probably what happened to your LM3886 if it isn't cooked already.
2) The MyRef works best on 8 ohm loads, not 4 ohm loads. For 4 ohm loads, you'll need a 20-0-20 trafo.
3) The matching of the Howland resistors is pretty important for stability and phase margin - no idea how closely the resistors are matched on the EBay kits, but I always match them to better than +/- 0.1%.
Thanks linuxguru. Sorry to sound dumb, I've never understood the term "rails" (me being new to the dual supply concept).
The datasheet says (quote) "Wide Supply Range 20V - 94V". How do we arrive at the 38v figure and that 24v is ideal and 28v is high?
But so glad you mention this because I did measure something like 39V on the PCB somewhere. I have several LM3886 spares. But so far doesn't look cooked because everytime I turn it ony for ~3-4 secs for test and it starts getting warm. I could be wrong though.
It was so difficult to get the toroidal here. A local professional winder did it for me and I paid 3k INR. I hope there is a way to get a voltage drop of 4 volts. Planning to insert a 100W bulb in series to each 28v tapping end. May be that will bring the voltage down a little?
The datasheet says (quote) "Wide Supply Range 20V - 94V". How do we arrive at the 38v figure and that 24v is ideal and 28v is high?
But so glad you mention this because I did measure something like 39V on the PCB somewhere. I have several LM3886 spares. But so far doesn't look cooked because everytime I turn it ony for ~3-4 secs for test and it starts getting warm. I could be wrong though.
It was so difficult to get the toroidal here. A local professional winder did it for me and I paid 3k INR. I hope there is a way to get a voltage drop of 4 volts. Planning to insert a 100W bulb in series to each 28v tapping end. May be that will bring the voltage down a little?
Last edited:
94V is +/- 47V, which is the absolute max rating before device breakdown on the LM3886 die. You can't drive any realistic load at the absolute max supply rating - anything above +/-35V is pushing it, and it's known from experience that the LM3886 runs best between +/-28V to +/-32V with 8 ohm loads.
You can try using the 100W bulb in series on the primary side - no idea how much the secondary voltage will drop, but it's worth checking out. It's the same principle used in the bulb testing units.
Thanks so much. I was though planning to put that 100w in series on the secondary. But Primary sounds good too. (Should I be expecting any inducted resonance etc.?)
28V is 7 volts below 35. Wondering why it should be a problem. Should I replace the ICs once I have a +/- 24v supply?
28V is 7 volts below 35. Wondering why it should be a problem. Should I replace the ICs once I have a +/- 24v supply?
I think you are mixing the 28VAC and the max voltage of it.
If the trafo charges a cap, you get the max voltage not the 28VAC.
To get the max voltage, have to multiply it by sqare(2) so 28V*1.41=39.5V. That is the 39 you saw. And that is really close to the max +-42V that is suggested for any long term usage. Plus usually you get (at least I get) more voltage out from the trafo if not loaded.
So you can have over 40V at the capacitors. Hope someone can explain this well 😱
From 24V supply you get about 33-34V. If you check the lm3886 doc, fig 35, for 4 ohm load 35V is the mentioned max. If you go over it, then the chip overheats at heavy load.
If the trafo charges a cap, you get the max voltage not the 28VAC.
To get the max voltage, have to multiply it by sqare(2) so 28V*1.41=39.5V. That is the 39 you saw. And that is really close to the max +-42V that is suggested for any long term usage. Plus usually you get (at least I get) more voltage out from the trafo if not loaded.
So you can have over 40V at the capacitors. Hope someone can explain this well 😱
From 24V supply you get about 33-34V. If you check the lm3886 doc, fig 35, for 4 ohm load 35V is the mentioned max. If you go over it, then the chip overheats at heavy load.
Last edited: