It depends on whether the top of the chip or the PCB is going to be hotter. The hotter side should have thermal paste (less than 0.005") instead of a pad.
opc: You finally found a use for paralleled LME49600s! That's great. 
You have probably seen this already - a good read on paralleled BUF634s.
http://www.ti.com/lit/an/sboa065/sboa065.pdf
I've never found a good writeup on the internals of the LME49600 but I'm assuming they are fundamentally similar. Points out why balancing resistors are not needed on the outputs:
"The output resistance of the BUF634 is about 10 ohms. Therefore, series output resistors for decoupling the individual buffers are no longer necessary."
They calculate the balancing currents moving between 3 of the buffers in figure 2, given typical offsets, to be between 3mA and 10mA.
Interesting that they also point out for video, using 3 in parallel to get 500mA (instead of two) is actually recommended due to increased output resistance at high frequency and power levels. Shouldn't have any effect at audio frequencies through.
You have probably seen this already - a good read on paralleled BUF634s.
http://www.ti.com/lit/an/sboa065/sboa065.pdf
I've never found a good writeup on the internals of the LME49600 but I'm assuming they are fundamentally similar. Points out why balancing resistors are not needed on the outputs:
"The output resistance of the BUF634 is about 10 ohms. Therefore, series output resistors for decoupling the individual buffers are no longer necessary."
They calculate the balancing currents moving between 3 of the buffers in figure 2, given typical offsets, to be between 3mA and 10mA.
Interesting that they also point out for video, using 3 in parallel to get 500mA (instead of two) is actually recommended due to increased output resistance at high frequency and power levels. Shouldn't have any effect at audio frequencies through.
Last edited:
Rather OT I realise but - they're showing no decoupling at all on those paralleled BUF634s
Also the justification given for paralleling 3 of them does not make a lot of sense. And since when has a single opamp had its output on pin5? Looks rather like sloppy, unchecked work to me - don't take it as gospel 
agdr:
Good info in that paper! I hadn't seen it before.
It is important to note that the output impedance stated there is not the actual output impedance of the finished amplifier when the buffer is place inside the feedback loop of the op-amp. In that scenario, the output impedance should essentially be zero, or very close to it.
I'll measure it sometime this week just for kicks.
john dozier:
I honestly don't know... if I send out boards for another project in the future, then I can always tack another run of these onto that order, but to make a run of just these boards by themselves would be extremely expensive and would probably drive the cost up to $50 per board.
I would say that there's a pretty good chance of seeing another 50 boards up sometime in the next year, but I can't make any promises at this point.
Everyone else:
I've attached a final list of everyone who managed to get boards to this post. I will be sending out payment requests in the next few minutes so please use this list to get your total.
Shipping is going to be $5.00 flat rate, but if you already have a pending headphone or amplifier order, then I've put $0 in the shipping column since I'll just toss the boards in with your other parts.
Cheers,
Owen
Good info in that paper! I hadn't seen it before.
It is important to note that the output impedance stated there is not the actual output impedance of the finished amplifier when the buffer is place inside the feedback loop of the op-amp. In that scenario, the output impedance should essentially be zero, or very close to it.
I'll measure it sometime this week just for kicks.
john dozier:
I honestly don't know... if I send out boards for another project in the future, then I can always tack another run of these onto that order, but to make a run of just these boards by themselves would be extremely expensive and would probably drive the cost up to $50 per board.
I would say that there's a pretty good chance of seeing another 50 boards up sometime in the next year, but I can't make any promises at this point.
Everyone else:
I've attached a final list of everyone who managed to get boards to this post. I will be sending out payment requests in the next few minutes so please use this list to get your total.
Shipping is going to be $5.00 flat rate, but if you already have a pending headphone or amplifier order, then I've put $0 in the shipping column since I'll just toss the boards in with your other parts.
Cheers,
Owen
No drawbacks per se, but if you're only going to populate one or two buffers you're much better off going with a normal BAL-SE version of The Wire. This board is really meant as a power amplifier for low impedance loads.
If you're driving a 600 ohm load, you will always be voltage limited and additional buffers will do absolutely nothing for you. The most current you can get with 18V rails and a 600 ohm load is going to be roughly 18mA RMS and a single buffer can supply 10 times that amount.
What you really want for maximum drive into a 600 ohm load is the BAL-BAL version which has twice the voltage swing.
Regards,
Owen
If you're driving a 600 ohm load, you will always be voltage limited and additional buffers will do absolutely nothing for you. The most current you can get with 18V rails and a 600 ohm load is going to be roughly 18mA RMS and a single buffer can supply 10 times that amount.
What you really want for maximum drive into a 600 ohm load is the BAL-BAL version which has twice the voltage swing.
Regards,
Owen
After the BAL-BAL comes agdr's suggested 'rework' that gets the rail voltage to 20v by using lme49860+49610. Virtually equivalent to the BAL-BAL, but with more noise, would be a pair of BAL-BALM (outlined in my blog) but that requires two SE-SE pcbs (rework will work). SE-BRGD using lme49610 could REALLY swing
http://www.diyaudio.com/forums/blogs/wolfsin/810-deeper-into-bal-balm-mono-version-bal-bal.html
http://www.diyaudio.com/forums/blogs/wolfsin/810-deeper-into-bal-balm-mono-version-bal-bal.html
I know that the bal-bal would have been best but I missed the boat on that, this LPUHP should fit the bill as most Sextette owners are resoting to old vintage receivers to drive these classic AKG's, Plus I have a pair of high effeciency full range fostex speakers that I could use this with (thinking I buy a few buffers a month as a digikey payment plan
. The lack of distortion with 10Vrms from the LUPUHP looks fine for these phones and at 600 ohms noise isn't an issue. My concern was just the unpolutated buffer traces turning into antennae.Wolfsin, when I looked at the Oveture datasheets on digikey it looked like the 20V to 22V models offer any no advantage over the 18V LME49990 used in the LPUHP as far as actual Vrms output before the thd spike. Maybe I am missing something but I think asking a low noise low distortion opamp to deliver more than 10 clean Vrms is to much, it seems to be the cutoff regardless the rail voltage.
Thanks, I will not use the 610 then. My own application uses 12v battery rails so only bridging gets me where I want to be. I appreciate not having one more alternative to try! agdr may be disappointed however.
Hi Guys,
For those who might be interested in a cnc machined chassis, would the preference be monoblocks or "duoblocks"?
Also, would it be best to make allowance for transformers up to 100va? This could make the size quite a bit larger and hence more expensive.
Haven't had time to do draw up any ideas yet, and I am open to suggestions re layout.
For those who might be interested in a cnc machined chassis, would the preference be monoblocks or "duoblocks"?
Also, would it be best to make allowance for transformers up to 100va? This could make the size quite a bit larger and hence more expensive.
Haven't had time to do draw up any ideas yet, and I am open to suggestions re layout.
My first preference is probably for monoblocks, but I will still be in either way if 'duoblock' proves more popular.
Duoblock option may actually give better options for arrangement of power supply cabling separation from signal cabling. I'm thinking fused IEC socket in the middle of back panel. Given the dimensions of the boards the IEC socket may dominate the rear of the enclosure with a roughly 30mmsq cut out for a fused IEC socket. Even more so in the case of monoblock enclosures.
Duoblock option may actually give better options for arrangement of power supply cabling separation from signal cabling. I'm thinking fused IEC socket in the middle of back panel. Given the dimensions of the boards the IEC socket may dominate the rear of the enclosure with a roughly 30mmsq cut out for a fused IEC socket. Even more so in the case of monoblock enclosures.
could it be designed so that you could do either via an interlocking system? just a thought. i would probably go for monoblocks i guess; as it is its breaking the system i had planned for my other amps, which is to have a central power supply with control circuitry for the 2 'mono' units, so a 3 box build for dual mono, but given the onboard power supply of the LPUHP this isnt really possible. i do like the idea of a stereo unit as it would simplify input connection to a single 4 pin, but mono makes it easier to place the units closer to the speaker
so i guess overall mono would be preferred.
so i guess overall mono would be preferred.