I am now working on a 500W at 8 ohms, 1000W into 4 ohms. Hitachi's, could work but I have a pretty good approach.
> Hitachi's, could work but I have a pretty good approach.
" but " ?
Hitachi's with clever heatsinking ...
" but " ?
Hitachi's with clever heatsinking ...
Not that it matters, but the die of the K1058 is identical to the one of the K135.
Lowest Rds(on) value of a K135 is 0.8 ohm.
Upside of the old 125W TO-3 Hitachi devices was 8 amps max Id instead of 7, downside was a much higher Rds(on)
(you can have my memory bank for a buck, the fella sitting in the dark corner of it comes complementary)
Yes.
John, you could have a look to Semelab ALF08NP20V5 (N&P in the same die) if 200V is enough for you.
Last edited:
For the increase of Rds(on) with rising junction temperature, at 125C, think in the order of double the number at 25C.
Interesting that Semelab prefers to rather concern themselves with Rds(on), instead of transconductance. (Must be a lateral thinking thing
)
No, thank you. I have already paid money to produce this double-sided PCB just for this test. Enough. The part in the same PCB and under same test conditions is deeply inferior to other opamps and it is enough for me not to use it. I do not care about 0.00000000000001% THD number, this parameter is almost useless.
Second section of the opamp in my test circuit has gain 40dB and even here the 4562 is worse than 5532, re EMI pick up. The case is closed, for me. If I need low noise low distortion BJT opamp, I go for AD797, not 4562 TI family.
Wrong product. In order to remove dirt through magic, buy Buybees. AD797 is only good when your plates are shiny and clean and you don't want to add nasties.
😎
In my recent opamp comparison, I like the sound of JRC2068 in inverting gain stage circuit (LF353 0dB --> 2068 6dB+ --> 2068 0dB). Not expensive but it is BIPOLAR so now becomes one of my favorite.
2068 is not popular in this site and I didn't think it was of good quality, but I was amazed (at least with specified PCB compared with other opamps) after I pulled one out of my Harman/Kardon receiver which has a lot of it (so I pulled from circuit that I wont need).
ADD: If with other opamps I tend to turn down the volume, with 2068 I tend to turn the volume up. I guess this is the sign of low distortion? But the vocal and soundstage is very nice.
Last edited:
2068's were very common in really budget gear of the late 80' and 90's. Budget as in sub £100 amp/radio/cd/phono +speaker systems. Often the SIL package. Mind you, that doesn't detract from the device itself. I've used them in place of 5532's before now.
I still don't get it, you say you don't care about a 0.00000000000001% THD number, but then you care about the noise results in an op amp configuration (high noise gain) that's never going to happen in any sane audio design? Sounds like a little compulsive, isn't it?
Then why did you spent money on new 2 layers layout (with through hole parts, as Mr. Marce noted above)? I've suggested myself a few pages up a 4 layer layout with ground planes and SMD parts, if you really want to investigate the intrinsic characteristics of the devices under test. Otherwise, it appears to me that the results you got (scaling down the noise) may very well tell "the LM4562 is a more demanding op amp when it comes to layout than its XYZ competitor in this test". Which is a valid and important point for DIYers, not really happy to handle 0802 parts, if they consider the performance metric you choose (noise at high noise gain) makes any sense for audio. But it still doesn't tell much about the op amp intrinsic performance.
Really? I can't remember that. Mostly I saw 45xx series or 5812.
From the many 2068s in the H/K receiver, only one is different, it is 4556. It is a very high voltage gain amp, even at low supply rail. May want to "hear" it's sound someday.
Just curious Mr. Curl, if it's not a trade secret, at what current do you bias such an 18 pair amplifier? It it's only 50mA per pair, that would be a 450mA, or with +/-90W rails ( for 400W/8ohm, needs 160Vpp) that would be a 81W/channel dissipation at idle. The total worst case dissipation (at about 1/3 max output power into 4ohm is 81W (idle) + 350W (worst case power dissipation) = 431W per channel, or 862W per amp. These are amazing numbers from a thermal management perspective, how do you manage this? Are you using any sort of forced cooling?
Ok, so the idle power dissipation is in fact about 250W (1.35A*90V*2). Which adds up to 600W worst case dissipation for one channel (I suppose these are mono blocks, otherwise 1.2kW worst case dissipation for stereo is... humongous). The question remains, how do you take out those 600W out of the heat sink, to keep the junctions to <150 degrees? You would need a heat sink with better than 0.1C/W for a 90 degrees heat sink temperature, that's a junction temperature of about 140 degrees if we account for the case to junction and the case (isolated) to heat sink thermal resistances.
0.1C/W is about 24"x12" of aluminum profile with 1.5" high, 0.5" spaced, fins (ok, 25% smaller if it's black anodized, also 10% smaller because the heat sink to air thermal resistance decreases with the heat sink temperature) placed in an optimum position (no air obstruction). Is this what you are using?
Or simply don't you guarantee your amp for the worst case thermal condition and shut it down when a certain heat sink temperature is exceeded?
yes, more or less very close from each others between all lateral power devices I used.(Must be a lateral thinking thing
Expect to lose in practice around 5V between driver and power devices output.
But ... i had seen their transconductance's curves in their data sheets as well.
Last edited:
It was more an academic thought and also a little positive promotion for SMD designs in audio... There does seem to be some bad press for SMD in some walks of audio, yet it is used in other areas of analogue (some audio included, but not high end audiophile stuff, just professional and communications).
Nice test and interesting results anyway cheers for doing it.
- Status
- Not open for further replies.