Hi all,
I'm building a 3-way active system at the moment and using a whole pile of LM3886 for the mids and tweeters. While they're good, I know that one can build something better and I'd like to do that as a next step in upgrading it, at least for the front speakers.
So: do you have recommendations for solid-state amps with the following sort of characteristics:
- Class AB[1]
- about 50W RMS into 8R
- can run from 35+35V DC rails (I have 25+25 transformers that I will reuse)
- I would like to avoid boosted+regulated input+VAS supply but will do it if necessary, just not sure where the extra hardware will go yet...
- the best possible (quantifiable!) performance, particularly at higher frequencies
- integrated turn-on mute and DC protection if possible[2]
- at most 100mm of heatsink length per channel[3], so a small count of high current transistors might be good
- don't worry about the LF amps, that bit is easy[4]
Should I just make a (variant of) Blameless or Low-TIM ? They're both pretty old designs and I haven't been paying any attention to amplifier design in the last 15 years; have they been significantly improved-upon?
Links to schematics: good. Links to PCBs and/or kits: great. Either way, something well-proven from measurement and not just simulation. I can do custom PCB layouts to meet the 100mm requirement but don't want to be tweaking schematics of proven designs and probably making them worse.
thanks...
[1] I realise that Class A is kind of an obvious thing to consider here, but space and power requirements forbid it. I have many channels and limited physical room (rack cases) to house the amps and really don't want a rackfull of class As hoovering away, heating up the lounge room.
[2] I intend to directly connect midrange and tweeter drivers to these amps and would like to have the amplifier design be safe enough in turn-on/off behaviour that I can avoid any series protection capacitors, etc.
[3] again with the rack cases and heatsink fan tunnels. Heat dissipation capacity is kind of huge but the PCB for each amp needs to be about 100mm long so that I can physically put channels in a single case (with the LF amps!) to replace the 3886s.
[4] I have plenty of good designs in the 100-150W RMS range to choose from to run the LF drivers. It's a rare modern amp design that doesn't have incredibly good performance for all frequencies below 500Hz...
I'm building a 3-way active system at the moment and using a whole pile of LM3886 for the mids and tweeters. While they're good, I know that one can build something better and I'd like to do that as a next step in upgrading it, at least for the front speakers.
So: do you have recommendations for solid-state amps with the following sort of characteristics:
- Class AB[1]
- about 50W RMS into 8R
- can run from 35+35V DC rails (I have 25+25 transformers that I will reuse)
- I would like to avoid boosted+regulated input+VAS supply but will do it if necessary, just not sure where the extra hardware will go yet...
- the best possible (quantifiable!) performance, particularly at higher frequencies
- integrated turn-on mute and DC protection if possible[2]
- at most 100mm of heatsink length per channel[3], so a small count of high current transistors might be good
- don't worry about the LF amps, that bit is easy[4]
Should I just make a (variant of) Blameless or Low-TIM ? They're both pretty old designs and I haven't been paying any attention to amplifier design in the last 15 years; have they been significantly improved-upon?
Links to schematics: good. Links to PCBs and/or kits: great. Either way, something well-proven from measurement and not just simulation. I can do custom PCB layouts to meet the 100mm requirement but don't want to be tweaking schematics of proven designs and probably making them worse.
thanks...
[1] I realise that Class A is kind of an obvious thing to consider here, but space and power requirements forbid it. I have many channels and limited physical room (rack cases) to house the amps and really don't want a rackfull of class As hoovering away, heating up the lounge room.
[2] I intend to directly connect midrange and tweeter drivers to these amps and would like to have the amplifier design be safe enough in turn-on/off behaviour that I can avoid any series protection capacitors, etc.
[3] again with the rack cases and heatsink fan tunnels. Heat dissipation capacity is kind of huge but the PCB for each amp needs to be about 100mm long so that I can physically put channels in a single case (with the LF amps!) to replace the 3886s.
[4] I have plenty of good designs in the 100-150W RMS range to choose from to run the LF drivers. It's a rare modern amp design that doesn't have incredibly good performance for all frequencies below 500Hz...
Perhaps a VSSA variant? I've built a few and find them to be a really nice sounding amplifier well suited to your operating parameters. There are several threads related to these little amps, one of them my own (shameless plug).
TDA1514 might fit the bill - looks to me to have more potential than LM3886, despite still being a chipamp. Needs to be bridged to hit your 50W/8R target.
I suggest VSSA variant too. It is much better than LM3886 or other amplifier with more complex design (subjectively) in high frequency.
By VSSA, you mean one of these? Frankly it looks way too simple to be particularly good; has anyone measured a built VSSA to show that its THD+IMD+N is in the same ballpark as Blameless and similar? If true, that would be really awesome.
Has anyone built a "mini badger" low-power version of the honey badger? I'm looking for that level of quality but obviously don't need nearly as much silicon on the output.
The TDA1514 specs for THD+N look very good. I'd have to rebuild my power supplies though and it will do only 28W into 8R. The -3dB probably doesn't really matter but buying new transformers is annoying/expensive, and I don't see the chips listed at Mouser or Farnell.
Has anyone built a "mini badger" low-power version of the honey badger? I'm looking for that level of quality but obviously don't need nearly as much silicon on the output.
The TDA1514 specs for THD+N look very good. I'd have to rebuild my power supplies though and it will do only 28W into 8R. The -3dB probably doesn't really matter but buying new transformers is annoying/expensive, and I don't see the chips listed at Mouser or Farnell.
I reckon you can use your existing 25+25 trafo to give a single ended supply, suitable for running two bridged TDA1514s. This means your smoothing caps going in parallel rather than in series.
You're right the chips aren't current production but the secondary market has plenty.
You're right the chips aren't current production but the secondary market has plenty.
That is the basis for what I am talking about. It exists in a few different configurations and implementations. To the best of my knowledge those who have built one have all spoken very highly of it. This includes myself; I have built five slightly different variants and not been disappointed by one yet. Here are a few examples of VSSA builds:
http://www.diyaudio.com/forums/solid-state/231662-peeceebee.html
http://www.diyaudio.com/forums/solid-state/237219-fet-hex-explendit-amplifier.html
http://www.diyaudio.com/forums/solid-state/244903-vssa-through-hole-pcb-build-thread.html
And my own most recent efforts:
http://www.diyaudio.com/forums/solid-state/250097-vssa-through-hole-version-jason.html
It is a relatively high speed design and, as an example, has demonstrated near perfect square wave tests in excess of 100kHz. Sorry, for myself I do not posses the equipment to make meaningful measurements of my VSSA builds with respect to THD, IMD and other 'baseline' specifications. All I can say is that no matter what I have built and / or listened to in the place of the VSSA I end up going back to one of the VSSA based designs.
For its relative simplicity it is a really nice sounding circuit in its various incarnations and forms. The VSSA has a very articulate and natural delivery that makes it seem to 'vanish'. There is no 'signature' for this amplifier, it just does exactly what an amplifier should do. A quick example is playing very familiar music and hearing the lyrics so clearly that I realized for my whole life I had misheard the song! I'm not talking about one piece of music either, I have experienced this with different artists and genres. Hearing not just an instrument but even the details of the musician's technique that had not been noticeable before with other amplifiers.
I used to be a specification junkie. I still think there is merit in using specifications to make basic judgements about an unknown. Ultimately however the 'proof is in the pudding' and one has to use the experience of a particular amplifier's performance to decide if it pleases you. It isn't an enormous commitment to try a VSSA out and you may find yourself pleasantly surprised if you give it a try.
Thanks for those links, Jason. I take it that all of your excess boards have been sold? I see in your thread that you're considering a different CCS option - any chance you'll be spinning another batch of boards with that change? I'd also consider having the boards made myself if you're willing to make the design files public.
And where does one buy 2SK1058/2SJ162? Mouser/Farnell know nothing of them but I can buy locally for $20!!! each; are there any good/cheap online stores or trusted eBay vendors? Has anyone tried the Exicon ECX10N20/10P20 devices? Mouser lists a bunch of Fairchild+Infineon 10N20 devices; has anyone tried those? $1 for FQU10N20CTU but I don't see any complementaries there.
If I could get a cheap pair of FETs I'd probably veroboard this to have a play, not that I have any test equipment beyond an analogue CRO to validate the design with. It's not going to happen at $40/pair though.
And where does one buy 2SK1058/2SJ162? Mouser/Farnell know nothing of them but I can buy locally for $20!!! each; are there any good/cheap online stores or trusted eBay vendors? Has anyone tried the Exicon ECX10N20/10P20 devices? Mouser lists a bunch of Fairchild+Infineon 10N20 devices; has anyone tried those? $1 for FQU10N20CTU but I don't see any complementaries there.
If I could get a cheap pair of FETs I'd probably veroboard this to have a play, not that I have any test equipment beyond an analogue CRO to validate the design with. It's not going to happen at $40/pair though.
My suggestion is to stay with something you are, or will be, familiar with, LM3886 amps, get the speakers working properly THEN think about a different amp for the tweeters. My bet is you won't bother because the effort for (any) return may not be worth it.
Abs
ps I don't have Golden Ears, just normal human ones and I also know my senses can be deceived by any number of things.
Abs
ps I don't have Golden Ears, just normal human ones and I also know my senses can be deceived by any number of things.
Last edited:
TSSA 1.7 being being sold by Sonnya in the Vendors section. SITO Audio. Fantastic airy highs. Not harsh at all.
I don't work super fast, so a re-spin may take a little time. The CCS thing isn't really an issue in my opinion; my PeeCeeBee version with no CCS sounds just as sweet as one with a CCS in any form I've tried. As for the current boards, I have assembled a stereo pair that are my test mules and have one unstuffed set I am planning to build with slightly different components as my 'production' build. That said there are other members that have expressed interest so simply re-ordering the existing design isn't out of the question either. I would give consideration to putting up a zip of gerbers.
For the basic design the MOSFETs must be laterals, I've used Tayda Electronics to get Renesas 2SK1058 / 2SJ162 devices for $5 USD each. The Fairchild parts are verticals and won't be appropriate for my version though users have built with BJT and Vertical MOSFET output stages. Suitable devices:
Renesas: 2SJ162 / 2SK1058
Semelab (Alfet / Magnatec): BUZ90xP series or ALFxxNxxV / ALFxxPxxV (orW suffix) series
Exicon: ECX / ECW series
Just be mindful of the package, I chose to use TO-247 / TO-3P devices and TO-264 won't fit my current design. This can be easily done on veroboard.
Laplace,
have you looked at and considered the driver sensitivities?
How much voltage do the three different speaker drivers require?
Have you looked at speaker protection relays and at ESP's site for auto Turn ON/OFF ?
have you looked at and considered the driver sensitivities?
How much voltage do the three different speaker drivers require?
Have you looked at speaker protection relays and at ESP's site for auto Turn ON/OFF ?
Oh, absolutely. I'm currently assembling the first stereo pair (6 amps in a 3U case) and it will probably be 6mo before I have all three of those modules completed. Once I'm used to it and have the crossovers tuned I'll think about replacing the 3886s for the front pair, maybe early 2015. This thread is me trying to catch up on where amplifier design has been going recently because I haven't paid it any attention in over a decade. The whole *SSA thing is new to me for example.
I may decide to leave the power amps alone and upgrade the DCX analogue boards instead.
That's no problem at all; as above...
wow, that's cheap, and motor-control FETs too! I'll have to do some datasheet reading and see if their shipping is affordable.
thanks!
Thanks, I'll have a look into that too.
re The First One; I read that thread and can probably believe that it's a lovely amplifier but I'm not interested in buying prebuilt modules or spending that much money, especially if I could build a slightly different VSSA for 15-20% of the price. Remember, I will need four of them to upgrade my stereo pair; six if I include the LF. I ain't poor but I'm not about to plunk down EUR500+ for this either.
Yeah, the mids & tweets have a couple dB more sensitivity than the woofers; my current build is 100+50+50W/ch on the pink-noise spectral assumption of similar power per decade (30-300, 300-3k, 3k-20k+ are my approx intended starting points). I don't reckon I'll be anywhere near the power limits before my head explodes though.
I'm trying to avoid separate output protection relays and go with inherent muting & current limiting where possible. If nothing else, I'd struggle to fit any additional boards+relays in there!
Last edited:
Take the same circuit as for your mids and reduce the power supply, or use the extra overhead to regulate the IPS/VAS. Drop to two pairs of outputs.
In my learning, it turns out it does take a lot more oomph in the treble than I thought. I don't ever want to clip my tweeter amp. I sure don't want to have to put in I/V limiting. At least I am protected against DC. I may build it's peak output to be only about the same as the main amp, and then scrimp on the heat sink and size of the power supply.
In my learning, it turns out it does take a lot more oomph in the treble than I thought. I don't ever want to clip my tweeter amp. I sure don't want to have to put in I/V limiting. At least I am protected against DC. I may build it's peak output to be only about the same as the main amp, and then scrimp on the heat sink and size of the power supply.
Member
Joined 2009
Paid Member
I understand you very well. It is not cheap but the output devices alone is 19% the price tag. I am not overcharging.
Go for the Eliot design (TGM8) but replace the output devices with NJW0281 and NJW0302. They are a good replacement and they will only cost you 3$ each.
All capacitors should be something like panasonic FC caps..
With this you are down to just short of 20€ per channel excluding PCB and heatsinks.
- Sonny
- Status
- Not open for further replies.