SIP-8 to SOIC-8 adapter

dotneck335

Member
2012-12-11 10:10 pm
I am trying to replace a HA12017 IC with a modern SOIC chip. The Hitatchi chip is a single SIP-8 op amp (pin 1 Out, pins 6 & 7 In, pin 4 -V, pin 8 +V), and I need a suitable adapter to say, an OPA1641. I'm having difficulty finding one----???? The usual sources all have ones that are pin-to-pin, but that won't work in this case, as the OPA1641 (and most single SOICs) are pins 2 & 3 In, pin 6 Out, pin 7 +V, pin 4 -V).
 
Last edited:
HA12017 isn't just any old opamp. It's a very low voltage noise (possibly <1 nV/√(Hz)), externally compensated (2nd order no less!) multistage dedicated preamp IC capable of substantial output voltage swings with very low distortion for the day. I don't think I've ever seen it in anything but MM phono stages (the datasheet is also heavily geared towards this use) although I strongly suspect that it has rather too much current noise to be ideal for this application - for a low-noise bipolar input part I might expect as much as 1.5-2 pA/√(Hz). (I have some difficulty recreating the datasheet specs im sim, I suspect the equipment used may have tended to underread by 10% or so, or maybe I'd have to recreate the filter exactly as given instead of just integrating 20-20k.)

Not sure why they didn't market it towards MC phono inputs instead.

What are yours being used for? 8 of 'em sounds like studio equipment again.
 

Attachments

  • ha12017-datasheet-2.asc
    6.7 KB · Views: 22
Last edited:

dotneck335

Member
2012-12-11 10:10 pm
HA12017 isn't just any old opamp. It's a very low voltage noise (possibly <1 nV/√(Hz)), externally compensated (2nd order no less!) multistage dedicated preamp IC capable of substantial output voltage swings with very low distortion for the day......... I strongly suspect that it has rather too much current noise to be ideal for this application - for a low-noise bipolar input part I might expect as much as 1.5-2 pA/√(Hz). (I have some difficulty recreating the datasheet specs im sim....
What are yours being used for? 8 of 'em sounds like studio equipment again.
I am having difficulty figuring out how you came up with a voltage noise spec of 1nV√Hz from the datasheet---which says 5.3mV noise after a gain of 75.9 db, measured through an 'A' weighted filter. The "substantial voltage swings" is also puzzling, as the datasheet says 14.7 volts out at 0.1% THD with a supply of +/- 24 volts. Please explain.
 
If you simulate things like I did (the file I attached was for LTspice as usual), the noise spec for 43 ohm source impedance (1.15 mV typ after another +40 dB and A-weighting) plus the given circuit values will give you an idea of the required voltage noise spec, while the 3.3 kOhm spec (5.3 mV typ) gives you an idea of current noise levels. (Basically you try to replicate the setup and fiddle with opamp model e_n and i_n.) Plus, I very much doubt that they would have chosen 165 ohms in the ground leg if the part wasn't at least that good (1.6ish nV/√Hz).

14.7 V is the maximum output spec - that's 41.5 Vpp (out of 48). Then there's a spec with 10 V out at 1 kHz at 0.002%. That's a good +22 dBu for ya. Mind you, +/-24 V is higher than usual these days, it would have been less exotic in the late '70s.
 
Last edited:

dotneck335

Member
2012-12-11 10:10 pm
If the part (HA12017) was really that good (1.6nV√Hz), it would seem that it would have shown up in a LOT more circuits than the 3.5nV√Hz NE5534; but that doesn't seem to be the case. Here's the way I figured it: 5.3 mV noise out is measured through a IHF 'A' filter, which lowers the actual 'flat' reading by about 4 db. That makes it ~8.4mV. The amplification shown in the test circuit is 75.9 db, making the E.I.N. 1.347µV. Of that, the 3.3KΩ source resistor contributes ~ 1.04µV, and the opamp 0.898µV (using the square root of the sum of the individual noise sources each squared). 898nV of noise equates to 6.35nV√Hz. That, of course is both voltage noise AND current noise. I'm not sure how it's divided up, but in any case it's more than the NE5534's 3.5nV√Hz voltage noise and 0.4 pA√Hz current noise.
If the HA12017 will only deliver 14.7 volts max output with a +/-24 volt supply, how much will it put out with a more-normal +/-18 volts? I don't know, but it's surely a LOT less than an NE5534s 16 volts with that same supply.
 
Last edited:

dotneck335

Member
2012-12-11 10:10 pm
If the HA12017 will only deliver 14.7 volts max output with a +/-24 volt supply, how much will it put out with a more-normal +/-18 volts?
Thinking further on the subject........
You're right, the 14.7 volts is RMS, which makes for a 20.8 volt peak outta 24 volts; maybe 15.6 volts from a +/- 18 v supply; plenty enough.
Another troubling thing about the HA12017 is the abysmally poor -PSRR---it's only -45db! That's not too good.......also no slew rate quoted...hmmmm......
 
Abysmal? Actually, nope. That's 45 dB of PSRR in an RIAA amp circuit at 100 Hz, where closed-loop gain is +58 dB. To little surprise of anyone who has gone through Bruno Putzeys' presentation from several years back (the VAS is probably referenced to V-), both add up to the 103 dB of open-loop gain at this frequency.

There aren't too many audio opamps substantially better than this to this day - 103 dB of OLG at 100 Hz means you need a GBW of 14 MHz.

Since compensation is external, chances are that you could affect slew rate via choice of components there, especially if don't actually need unity-gain stability as the standard RIAA circuit would. Unfortunately the datasheet doesn't give very much in terms of hints, so those who have toyed with amplifier compensation before may be at an advantage here.
 

dotneck335

Member
2012-12-11 10:10 pm
Yes, I hadn't noticed at first that the spec included the RIAA LF boost. Interesting that the supply rejection is 11 db worse on the -V side, so they put an additional 470µF cap on the negative supply line in the application circuit.

The HA12017 circuit that I am attempting to improve (in an Otari 8-track tape machine) has a gain of ~ 30db.,They used a 22pF compensation cap from the output back to comp pin 5 and a 24KΩ resistor from -Vss to offset pin 3. I don't think I'll need either of those with my replacement NE5534.
 
Last edited:

dotneck335

Member
2012-12-11 10:10 pm
You wouldn't have any idea of tape head R and L in this one, would you? If this one is similar to an MM phono cartridge again, chances are that a 5534A would do well. Minimum gain should be high enough that no external compensation should be necessary (I think it's decompensated for a gain of something like 3). What's the feedback network like?
The tape head is ~ 650 mH, 650 Ω, 3.2 mV @1 KHz. The NE5534A is internally comp'ed for gain ≥ 3.
 

Attachments

  • Otari Repro PreAmp.pdf
    29.4 KB · Views: 22
So yep, squarely in MM territory.

Potential risks:
1. Gain returns to unity at a few MHz. May need inserting a resistor in series with 100p C105 (ca. 680 ohms) to keep minimum closed-loop gain high enough for stability. Seems the better-performing alternative vs. using a compensation cap.
2. No R in series with input whatsoever. ~200R would be good. But try as-is.

C102 presumably is a 47p, not a 47n.
 

Attachments

  • otari-repro-ne5534a.asc
    4.3 KB · Views: 11
Last edited: