You can't safely get 68W out of a MyRef for several reasons:
1) At +/- 32V rails (close to optimal for sonics), the LM3886 is not spec'ed to drive 4 ohm loads, only 8 ohm loads. You could however try 6 ohms at +/- 28V and hope that SPike doesn't get triggered.
2) There's a series resistance of 0.47 ohm in the Howland current pump that scales the output voltage swing into the load by the ratio RL/(0.47 + RL). This reduces the output power by the square of that ratio, making it progressively lower for lower RL, i.e. worse for 4 ohm loads than 8 ohm loads.
3) It's more stable at higher RL, both due to SPike as well as the phase margin from the Rev C compensation schema, which is optimized for 8 ohms.
You can reasonably expect ~40W max into 8 ohms with ~25V amplitude at the output at +/- 32V rails.
Does this reasoning mean you shouldn't use the Rev C to power 4 ohms speakers, even if at lower supply voltages?
Isn't that a huge limitation for a power amplifier?
Isn't that a huge limitation for a power amplifier?
No, just use a 2x22V trasformer and it runs fine at 4 Ohm (and 8 Ohmm too).
It's just that if you have 8Ohm speakers you can go a bit up with voltage (2x25V) to have a bit more power.
To talk about "a bit more power" is relative. What's the difference between 40 watts @ 8 ohms and 40 watts @ 4 ohms?
What I wonder is how dynamics are affected with heavier loads, like 4, 3 or 2 ohms, which is what may happen with real world loads.
That is: an 8 ohm speaker might dip lower in impedance, particularly at low frequencies, like 5 or 6 ohms.
What I wonder is how dynamics are affected with heavier loads, like 4, 3 or 2 ohms, which is what may happen with real world loads.
That is: an 8 ohm speaker might dip lower in impedance, particularly at low frequencies, like 5 or 6 ohms.
Mauro Penasa specced this amp at 22V for 4Ohm loads and at 25V for 8Ohm loads.
Loads intended as speaker manifacturers intend, dips are normal and the MyRef is perfectly able to handle them.
Quite all people built this amp is happy and from what I know the MyRef drives a variety of speakers with authority.
Obviously I don't thinks it's the best amp for an electrostatic speaker set...
Assuming 25V output amplitude for the rated max power, the feedback divider network of 12k and 390R gives ~786 mV amplitude at the opamp input. There's an input attenuator of with a ratio of 1/1.033, so that gives ~812 mV at the input connector block. Divide by ~1.4, giving ~575 mVrms - almost the same as your estimate (which may not have taken the input attenuator into account).
No static speakers for it, unfortunately. But I am working on a nearfield speaker project with 4 ohm drivers, so I was wondering if the MyRefC was up to the job. I was certain it was, and I was surprised at the output swing limits mentioned that I might have.
Let's see: 22v + 22v should be around +/- 29VDC, right? Can I expect around 55w @ 4 ohms? Of course I would never listen at that continuous power, but it's good to know I have some 6 or 9dB to swing up from 10 watts, which is a reasonable level for nearfield.
Well, I was revising the buffer question I had asked and found a bit of an incongruence there.
Of course I am asking if I would need a buffer for using a passive attenuator, a stepped or simple volume pot, even a balance control. What about an LDR attenuator? Can I connect it directly?
I was considering something simpler, like a 21-step, 10K attenuator you can get on eBay. But I wouldn't want to use a buffer, if possible.
Just search on eBay "DACT Type 21 Stepped Attenuator"... 10K value.
This little thing sounds incredibly good.
I'm still trying to make my DCB1 + LightSpeed attenuator sound at least as that attenuator alone!
This little thing sounds incredibly good.
I'm still trying to make my DCB1 + LightSpeed attenuator sound at least as that attenuator alone!
Last edited:
That's the "little thing" I was considering using for this project. It seems neat and very affordable. But I was concerned if I needed a buffer.
What's a DCB1?
Exactly that: a buffer, a very good one with no coupling caps, no feedback and a super voltage regulator by Salas.
I'm waiting from Hong Kong a second little stepped attenuator to try instead of the lightspeed.
So I hope to indentify between the DCB1 and lightspeed who's degrading sound compared to the attenuator.
OK, I didn't realize this. I went back to Mauro's website and managed to pick out that he spec'd 24 Vac secondaries, and rated the amplifier at 40 W at 8 ohms, and 56 W at 4 ohms. However, I can't determine which version this is.
Other the years, it seems that this circuit has been refined by others.
Can you please point me to a complete set of specifications for the latest generally available version?
Thanks and regards,
Rob
A few values of passives have been refined over the years, while most have remained unchanged. The functional specs are still the same, and the modified schematics are functionally equivalent to Mauro's Rev C schematics. Check out Dario's zipfile of documentation that he maintains and uploads here from time to time.
Obvious question: what buffer is that? Circuit please...
What Salas supply is that? Shunt certainly. Circuit please...
Tnx.
Wow, complicated project that one is!
Two thirds of the board are supply space!
Can't you power that with something simpler, like a good 3x7?
The relay controlled output is not something that I care too much about, as I will probably leave it on all the time.
Any idea of how much the matched 2SJ170BL pairs might cost?
If compared to the original Mauro's not a single value was changed, it was only added a bypass for a single cap (C9). So all data present on Mauro's site is still valid.
If compared to TP MyRef all PS bypasses (not present in the original) were removed.
The zip file was maintaned by RegiRegi, not me... I've made a final BOM and a tutorial.
The original Pass B1 is rather simple, it's the Salas PS that is complicated but it's way better than a LM317... 😎
Not much...
You can find them on the Swap Meet here on DIY Audio or on eBay
The main difference between the B1 and the DCB1 is that the latter has a split supply, which allows a no-cap output.
I'm pretty sure the Salas PS contributes a great deal to that buffer, but I also know that a 3X7 ps, properly filtered an bypassed, particularly in in the adj leg, goes a long way and shouldn't be disregarded. Much better than a 7XXX supply.
But I think Salas' ps is worth a try too.
If the DACT type attenuator sounded close to the DCB1 with Lightspeed, then it must be very good.
Dario,
I'm not quite understanding your setup. I have both of Uriaha's kits and they work great with the pots supplied. I thought the whole point of the LDR approach was to eliminate the need for the signal passing through a physical mechanism and eliminate the need for a buffer. If I remember correctly several people have said " the best buffer is no buffer at all", which adds to the purity and simplicity of the signal path. Having a Lightspeed or Lighter Note combined with a stepped attenuator sounds redundant to me. Can you clarify your approach?
Thanks, Bob M.
I'm not quite understanding your setup. I have both of Uriaha's kits and they work great with the pots supplied. I thought the whole point of the LDR approach was to eliminate the need for the signal passing through a physical mechanism and eliminate the need for a buffer. If I remember correctly several people have said " the best buffer is no buffer at all", which adds to the purity and simplicity of the signal path. Having a Lightspeed or Lighter Note combined with a stepped attenuator sounds redundant to me. Can you clarify your approach?
Thanks, Bob M.
Last edited:
- Status
- Not open for further replies.