Check the swap-meet section for the MiniRef boards being offered by PreSapian:
http://www.diyaudio.com/forums/swap-meet/254289-miniref-pcb-giveaway.html
Meanwhile, I verified that several fast opamps are quite happy and stable in my MiniRef 3886 - LT1361, LT1364, AD8022 - apart from the usual suspects mentioned earlier: NE5532, OPA2134, LME49720. YMMV, but just about any opamp can probably be stabilized with minimal compensation tweaks to the MiniRef 1875.
http://www.diyaudio.com/forums/swap-meet/254289-miniref-pcb-giveaway.html
Meanwhile, I verified that several fast opamps are quite happy and stable in my MiniRef 3886 - LT1361, LT1364, AD8022 - apart from the usual suspects mentioned earlier: NE5532, OPA2134, LME49720. YMMV, but just about any opamp can probably be stabilized with minimal compensation tweaks to the MiniRef 1875.
Just tried out the LME49720 and LT1364 in a MiniRef 1875 with C10/C20=54pF, C9/C18=220uF/4V BG-PK, C7=6.8nF, C8/C11=4700uF.
Result: Both opamps are stable with excellent sonics - I prefer the LT1364. Dark, detailed, airy.
Also tried the LF03d discrete opamp, and it worked for a while and then went into thermal runaway. I'd expect that C10/C20=100pF will stabilize it, but will try that later after testing out a bunch of monolithics.
Result: Both opamps are stable with excellent sonics - I prefer the LT1364. Dark, detailed, airy.
Also tried the LF03d discrete opamp, and it worked for a while and then went into thermal runaway. I'd expect that C10/C20=100pF will stabilize it, but will try that later after testing out a bunch of monolithics.
Actually, all the LM3886 models I found on the net (including the official NatSemi/TI models) were flaky. What I use is actually an LM1875 model posted by Pedja Rogic many years ago on a European audio forum - it turned out to be usable for the LM1875, perhaps a bit optimistic on the TH20 numbers, but fairly reliable (even conservative) for stability.
I used the same model as a proxy for the LM3886, but that turned out to be a mistake - the prototype MiniRef 3886 turned out to be unstable with the default values even though the sim (with the LM1875 model) showed adequate phase margin. However, I was able to stabilize the MiniRef 3886 with compensation tweaks and an additional RC network between the + and - inputs of the LM3886. I subsequently found that a similar RC network is probably needed (or doesn't hurt) for the MiniRef 1875 prototype also.
(I'll dig out the LM1875 model later - it's on another machine, not connected to the net.)
Link to a posting showing pic of the prototype MiniRef 1875 in a compact 1105 extruded aluminium cabinet, with 2x 19.5V, 2.1A netbook AC adapters for the rail supplies:
http://www.diyaudio.com/forums/chip-amps/79303-chip-amp-photo-gallery-273.html#post3919665
http://www.diyaudio.com/forums/chip-amps/79303-chip-amp-photo-gallery-273.html#post3919665
I am in the process of doing the same.
After reading a few "opamp compensation" papers I decided to adopt the simplest compensation for a non unity gain stable amplifier and went the noise gain route, i.e. adding and RC across the +IN to -IN pins.
For the 3886 I ended up with 390r + 10nF and it increased the HF noise gain to +30.8dB from the MF gain of +20.1dB, i.e. 10.7dB more noise at HF.
This was checked by adding load capacitance before the output L//R that is in the final assembly.
I have not yet connected speakers.
I have listened to noise via 40ohm headphones direct to amp output. Silence with shorted input <0.1mVac and slight noise with open input (Rin= 23k) ~0.2mVac.
I have yet to try an R+C across the feedback resistor.
Last edited:
The RC product is in the same ballpark as Mauro's 3.3k + 1nF and my 3.16k + 820pF (nothing special about it, just had that value PPS capacitor lying around). C=10nF is an order of magnitude higher, so I'd expect it to be even better at attenuating HF. No idea about the audible impact due to phase shift.
If you could do that, I would be very grateful.
I didn't know there was an official NatSemi model.
Which of the schematics on this thread have your 'compensation tweaks'?
(Replying to bcmbob here, to avoid cluttering the Chipamp Photo thread)
I have been thinking of switching power supplies for gainclones for a while now, mainly for the size and weight advantage over conventional 50/60 Hz mains transformers. Apparently, I am not alone - the use of a notebook AC adapter for the Class-A Amp Camp Amp by Nelson Pass confirmed and validated the approach.
I noticed (as he probably did) that the best values in switching AC adapters are clustered at either 12V or 19V - the latter is more useful for power amps, and you typically find 19V/3.42A adapters in wide use. However, they tend to be a bit larger than the ones I found, which are 19V/2A - perfect for each rail of the MiniRef. The downside is that I had to use two - one for each rail, thus doubling the physical volume. However, it all came together this week when I found the relatively compact netbook adapters shown - they fit comfortably within the 1105 case that I used.
I have been thinking of switching power supplies for gainclones for a while now, mainly for the size and weight advantage over conventional 50/60 Hz mains transformers. Apparently, I am not alone - the use of a notebook AC adapter for the Class-A Amp Camp Amp by Nelson Pass confirmed and validated the approach.
I noticed (as he probably did) that the best values in switching AC adapters are clustered at either 12V or 19V - the latter is more useful for power amps, and you typically find 19V/3.42A adapters in wide use. However, they tend to be a bit larger than the ones I found, which are 19V/2A - perfect for each rail of the MiniRef. The downside is that I had to use two - one for each rail, thus doubling the physical volume. However, it all came together this week when I found the relatively compact netbook adapters shown - they fit comfortably within the 1105 case that I used.
If you're using a 25-0-25 trafo, the DC rails are going to be about +/- 34V. So you're probably going to need 2x 36V, 5A SMPSes - not typically used in the notebook world, but often seen in industrial controls or telecom. These tend to be more expensive and better built than consumer-grade AC adapters for notebooks.
They'd be about the same size as the trafo + rectifier + filter caps, but weigh much less because of the use of ~250 kHz magnetics and electrolytics as opposed to 50/60Hz.
Price - it depends. I have access to consumer-grade 12..24V, 4A AC adapters at a retail price equivalent of USD 6 => I can build a +/- 24V, 4A supply for about USD 12. An 18-0-18, 4A trafo costs about USD 8 locally + the price of the bridge rectifier, snubbers and PSU caps (USD 4 at least). So it appears to be a wash: USD 12 in both cases.
50Hz trafo + bridge rectifier + caps is more reliable than an SMPS, until you short it (maybe by accident). That will invariably fry the bridge rectifier and maybe the trafo, while any halfway decent SMPS with foldback short-circuit protection will shut down and survive. In normal use without abuses like shorting or overload, a 50Hz trafo will probably survive for far longer than an SMPS with its stressed electrolytics and HV switching devices. However, you can improve the longevity of an SMPS by selectively upgrading a bunch of secondary-side electrolytics.
For the MiniRef 1875 with +/- 19V, 2A rails, in my build it seems fairly conclusive that 2x SMPSes do indeed take much less space than a trafo + rectifier + filter caps. There's no trafo available locally that fits within the ~1.25" height constraint of the case, but two netbook AC adapters fit comfortably side-by-side.
Here's the completed and fully wired MiniRef 1875 build, in a compact 1105 cabinet with 2x 19.5V/2A netbook AC adapters as the PSU:
As can be seen, everything fits within the approx. 208 x 115 x 48 mm cabinet.
The gold-plated stackable banana/binding-posts are my solution to the limited space available for fitting regular-sized binding posts to the cabinet. They're optional, of course.
An externally hosted image should be here but it was not working when we last tested it.
As can be seen, everything fits within the approx. 208 x 115 x 48 mm cabinet.
The gold-plated stackable banana/binding-posts are my solution to the limited space available for fitting regular-sized binding posts to the cabinet. They're optional, of course.
Last edited:
Those bananas work well with two-way cable insertion.
Yup - I notice that you're using similar low-profile banana sockets, but with gold-plated conductors and eyelet tags. I'll upgrade to those in a future build - this was a proof of concept, so I used whatever was available locally.
Another advantage of using SMPS adapters: absolutely no audible 50/60 Hz hum between tracks. The silence is uncanny if you've got used to a slight background hum (from EI or similar mains trafos) between tracks.
Another pic to illustrate just how small the MiniRef case is:
The case on the left houses a Symasym, and the small vertically upright one houses the MiniRef 1875 (about 4.5" tall). The Nokia phone gives some idea of the relative sizes.
An externally hosted image should be here but it was not working when we last tested it.
The case on the left houses a Symasym, and the small vertically upright one houses the MiniRef 1875 (about 4.5" tall). The Nokia phone gives some idea of the relative sizes.
Thanks, Jayadev. BTW, the audible sonics are very good, and there's a complete absence of 50 Hz hum with the SMPS adapters. I've got another pair of adapters to experiment with and reverse engineer the schematics (very simple flyback design with a UC3842 primary-side controller), so I will be able to clone them, if needed, in the foreseeable future (seems unnecessary for now, there's flood of these adapters available locally for sub-Rs.500 prices).
Since it's a pure flyback design, there's also room for me to mod them for maybe 24 V/1.7A, instead of 19V/2.1A, just by tweaking a couple of resistors in the voltage-sense network. That will allow the MiniRef to have a slightly higher peak power capability of maybe 2x 30W, at the cost of higher power dissipation at low swings (not really desirable or necessary, but maybe useful to have the additional headroom with the MiniRef 3886).
Since it's a pure flyback design, there's also room for me to mod them for maybe 24 V/1.7A, instead of 19V/2.1A, just by tweaking a couple of resistors in the voltage-sense network. That will allow the MiniRef to have a slightly higher peak power capability of maybe 2x 30W, at the cost of higher power dissipation at low swings (not really desirable or necessary, but maybe useful to have the additional headroom with the MiniRef 3886).