Hello everyone. I am looking for semantics for an amp that could drive low an low impedance ribbon speaker, say less than one ohm. I don't know if this is practical. The alternative is to build an ribbon with higher resistance using multiple strips of aluminium foil, which also has the benefit of higher efficience. The multiple stirips variant is however somewhat more complicated to produce.
I am curious if someone have tried both approaches and could brief me on the details, maybe one of them is supperior.
Regards, Björn
I am curious if someone have tried both approaches and could brief me on the details, maybe one of them is supperior.
Regards, Björn
Hi bjorn,
You have not indicated if this is a mostly full range ribbon or just for highs or something in between.
If for highs I would keep the single strip for better imaging and add 3ohms of quality series resistance (losing a further 12dB efficiency) so you can drive it from any 4 ohm amplifier, a nice easy load/phase. A medium size of amplifier should suffice.
If for wide range, I'd add strips to bring it up to 3 ohms - say 2 more and add a smaller resistor or experiment with none.
Personally I like higher impedances than this and would probably use a large amp with 3ohm resistor in series and live with a 6dB efficiency loss.
Don't expect 120dB. But for 90dB levels...
Cheers,
Greg
You have not indicated if this is a mostly full range ribbon or just for highs or something in between.
If for highs I would keep the single strip for better imaging and add 3ohms of quality series resistance (losing a further 12dB efficiency) so you can drive it from any 4 ohm amplifier, a nice easy load/phase. A medium size of amplifier should suffice.
If for wide range, I'd add strips to bring it up to 3 ohms - say 2 more and add a smaller resistor or experiment with none.
Personally I like higher impedances than this and would probably use a large amp with 3ohm resistor in series and live with a 6dB efficiency loss.
Don't expect 120dB. But for 90dB levels...
Cheers,
Greg
A lot of the mosfet designs posted in this forum will work with additional output devices.
Note though that this amp will typically have much lower supply rails than your 8 / 4 ohm version.
If you know the minimum impedance of the speaker you can work out the maximum current and configure your output stage to suit.
Cheers
Note though that this amp will typically have much lower supply rails than your 8 / 4 ohm version.
If you know the minimum impedance of the speaker you can work out the maximum current and configure your output stage to suit.
Cheers
It's a fullrange ribbon, i.e. from 300 Hz and up.
If I get you right it is possible to build an amp which is more or less like any other amp, just that you use lower supply voltage and for that reason the amp can handle the low impedance load?
Regards, Björn
If I get you right it is possible to build an amp which is more or less like any other amp, just that you use lower supply voltage and for that reason the amp can handle the low impedance load?
Regards, Björn
Hi Bjorn,
That's a fair take - delivering serious amps and low volts. Trouble is, optimised, it's not the same amp! It's not a typical design scenerio for a designer trying to offer value for 8 ohms! But we could go close.
Take myhigh power ampo, which has 4 pairs of high current MOSFETs in the output. Under normal circumstances they're unlikely to be requiredcto deliver more than 4A each. The cct is designed around that with limits here and there.
Remove those and drop the supply and they can deliver more but there are new limitations that emerge because the load is seriously different from the usual/expected.
Easy to adapt to however.
Cheers,
greg
That's a fair take - delivering serious amps and low volts. Trouble is, optimised, it's not the same amp! It's not a typical design scenerio for a designer trying to offer value for 8 ohms! But we could go close.
Take myhigh power ampo, which has 4 pairs of high current MOSFETs in the output. Under normal circumstances they're unlikely to be requiredcto deliver more than 4A each. The cct is designed around that with limits here and there.
Remove those and drop the supply and they can deliver more but there are new limitations that emerge because the load is seriously different from the usual/expected.
Easy to adapt to however.
Cheers,
greg
The amp doesn't have to be anything special, just choose the right power supply voltage (lower than usual) and have a sufficient amount of output devices and be very carefull where the currents go and where you apply the feedback, physically.
Elektor has published a design rather long time ago. I'm sure someone here can dig it up for you as inspiration.
It might be an idea to let the amp be of the type "voltage in <-> current out"
Elektor has published a design rather long time ago. I'm sure someone here can dig it up for you as inspiration.
It might be an idea to let the amp be of the type "voltage in <-> current out"
Try the Krell
I've made a Krell KSA 50 clone (see www link in sig line) with 4 pairs of output devices rather than the standard 2 pairs, and others have calculated that this will do very well down to 1 ohm and, while struggling below 1 ohm able (in theory) to put out over 41A continuous into 0.6 ohms resistive....though I ain't gonna try it and don't know if I would actually achieve this in practice.
As I omitted the output protection circuitry from my build I won't be testing this into less than 2 ohms, but it does seem very powerful and was a rather "less" complictated build for something of this power level.
You may want to post in the krell thread and see if a configuration of the Krell ksa50 amp would be suitable; maybe with certain bias settings you can get what you are looking for.
I've made a Krell KSA 50 clone (see www link in sig line) with 4 pairs of output devices rather than the standard 2 pairs, and others have calculated that this will do very well down to 1 ohm and, while struggling below 1 ohm able (in theory) to put out over 41A continuous into 0.6 ohms resistive....though I ain't gonna try it and don't know if I would actually achieve this in practice.
As I omitted the output protection circuitry from my build I won't be testing this into less than 2 ohms, but it does seem very powerful and was a rather "less" complictated build for something of this power level.
You may want to post in the krell thread and see if a configuration of the Krell ksa50 amp would be suitable; maybe with certain bias settings you can get what you are looking for.
Ok, thanks everyone, the Krell KSA 50 clone could maybe be something. I guess I need something straight forward since my experience with amps building is quiet limited. One question though, what exactly is an 'output device'?
maybe you should consider chip amps, completely integrated audio power amps, paralleling 2-3 LM4780 dual channel amps running from +/- 10-12 V requires only minor changes to the data sheet schematic
in fact a quick start approach would be to find a hackable ht amp that already has these wired up in a nice chassis and ps (ps V will be too high though)
you'll need low value power resistors to parallel the outputs and possibly a noise gain R-C network across each amp to allow you to reduce the gain below the min stable value for the chip amps
in fact a quick start approach would be to find a hackable ht amp that already has these wired up in a nice chassis and ps (ps V will be too high though)
you'll need low value power resistors to parallel the outputs and possibly a noise gain R-C network across each amp to allow you to reduce the gain below the min stable value for the chip amps
bjorn.lindberg said:
Ahhh... a transistor. Yeah, if you don't know this the Krell is probably too difficult, go with a couple of chip amps in parallel. Search "parallel" or "parallel" or P chipamp/gainclone.
Chip amps in parallel are unlikely to share the current well enough to stop going into protection mode even at low output levels.
If one amp gets too hot and shuts down then the others will quickly follow - one at a time - until they cool down, then the first one gets hot etc.
It would be far simpler to take an existing cct (many are on this forum) and layout and upgrade the output stage to take the extra current.
Because the voltage rails will be lower you can use high current FETS with very low Rds on. Note though the topology you select must have the P-channel FETS running off the positive rail and the N-channel FETS off the negative rail. This is necessary to avoid losses associated with the gate turn on voltage and this could be as high as 8-10 volts and if your rails are only 20 volts this is too much loss.
If you choose a quasi-complimentary cct then the positive rail FETS drive must be boosted via bootstrapping or seperate higher voltage.
Finally the cct you select should have a way of maintaining the DC offset to zero volts to avoid heavy currents at no signal.
If you need further explanation or help selecting and modifying a cct to suit drop me a line here or email.
Cheers
If one amp gets too hot and shuts down then the others will quickly follow - one at a time - until they cool down, then the first one gets hot etc.
It would be far simpler to take an existing cct (many are on this forum) and layout and upgrade the output stage to take the extra current.
Because the voltage rails will be lower you can use high current FETS with very low Rds on. Note though the topology you select must have the P-channel FETS running off the positive rail and the N-channel FETS off the negative rail. This is necessary to avoid losses associated with the gate turn on voltage and this could be as high as 8-10 volts and if your rails are only 20 volts this is too much loss.
If you choose a quasi-complimentary cct then the positive rail FETS drive must be boosted via bootstrapping or seperate higher voltage.
Finally the cct you select should have a way of maintaining the DC offset to zero volts to avoid heavy currents at no signal.
If you need further explanation or help selecting and modifying a cct to suit drop me a line here or email.
Cheers
quasi said:
strange, the data sheet shows how to parallel them, projects in the chip amps forums parallel them, ...
Parallel chips work fine. A 2-Channel STK4231-2 100W/2 @ 8 ohms each channel.
Parallel the 2 channels, and it makes an excellent 200W @ 4 ohms amplifier.
Paralleling chips works fine, but another approach, is to have a chip drive an output stage consisting of several pairs of parallelled transistors to drive a low ohm load.
Parallel the 2 channels, and it makes an excellent 200W @ 4 ohms amplifier.
Paralleling chips works fine, but another approach, is to have a chip drive an output stage consisting of several pairs of parallelled transistors to drive a low ohm load.
jcx said:
I don't know which chip-amp you are refering to, I would be grateful if you could post an example. Remember though that the speaker load discussed here will dip well below 1 ohm (well below) and very high currents will be involved. If paralleling involves the use of sharing resistors then a lot of power will be wasted.
EWorkshop1708 said:
The second approach would be better because of the currents involved. A sub 1 ohm load will draw 8 / 4 times the current of a traditional output load.
Cheers
National Semiconductor app note, with bridged and/or paralleled examples. Includes a mention 4 LM3886 paralleled for low impedance loads.
The LM4780 should be able to run on fairly low rails fine. To parallel these up you need to use a low value resistor at the outputs to insure current sharing.
With 3 parallel LM4780 paralleled using 20V rails you get about 140W into a 1ohm load, 200W into .67ohm. Should be well within thier thermal limits.
The LM4780 should be able to run on fairly low rails fine. To parallel these up you need to use a low value resistor at the outputs to insure current sharing.
With 3 parallel LM4780 paralleled using 20V rails you get about 140W into a 1ohm load, 200W into .67ohm. Should be well within thier thermal limits.
Tweeker said:
Yeah looks cool, thanks. The link provided actually shows 2 x 2 paralleled amps running bridged into 4 ohms minimum, not 4 amps in parallel. A sub 1 ohm load would need 4 times the amps if the logic is followed. Never the less it demonstrates the ease of paralleling the amps, so thanks again.
I have dug out an Elektor article of an amp for Ribbon Speakers. The article quotes an impedance of 0.4 ohms so the currents would be more than twice what I thought around 50 amps peak (assuming no losses anywhere). This is one time I would agree that speaker cable might make a difference.
Also amp cct described in the article is quite complex and includes a DC servo to maintain the output at zero volts to avoid the problem I described earlier. So for first-timers paralleling chip-amps would be the way to go.
Good Luck & Cheers
Cheers
Thanks everyone for the input. I think however that this is a bit too complicated for me. So I guess I'll just try to increase the impedance of the ribbon instead, i.e. making more strips. This would also increase the efficience, which is nice.
- Status
- Not open for further replies.