NE5532 NE5534 THD

A resource for technical documentation. Datasheets, application notes, instruction manuals, books and links to resources are found in the Document Library.
ricardo
Posts: 262
Joined: Sun Apr 10, 2011 7:24 am

NE5532 NE5534 THD

Post by ricardo »

Douglas Self did a small 10W amp in Elektor that used a zillion 5532s as the output stage to get 1 ppzillion THD.

As part of the exercise, he tested loadsa 5532s and was surprised to find differences in THD between makers.

The test circuit is inverting and my Jurassic (1980's) experience is that the original Mullard/Philips product would easily pass 10ppm THD20k 22dBu 600R in such circumstances. I'd expect to see stuff at 20k showing above 10ppm non-inverting but only slightly.

In the early 1990's when I got back into commercial electronics, I found even ST (Philips) 5532/4s from Thai factories were nowhere as consistent as the originals from Mullard, Southampton. I wasn't so interested in THD in 1990, but in noise. The originals in 1980 ALL met the tighter 5534AN low noise spec.

Sam Groner tested somewhat less makers but got similar results in his mammoth OPA shootout. Check out his excellent website.

The main disappointment from both of them was that TI 5532s were slightly worse than other makers.

But there are some stirrings that TI might have cleaned up their act which is good news. http://groupdiy.com/index.php?topic=56979

Edit by wk: Fixed GroupDIY url. (Text formatting was inside the url argument which broke it.) Also changed thread title to improve Google indexing.
________________

I can't remember how I got this pic. I think it was when Elektor was doing a promotion to get more subscribers but there is nothing of that sort on their webpage .. so not sure about copyright.

Related reading: viewtopic.php?f=6&t=1089
Attachments
5532distortion.pdf
(121.98 KiB) Downloaded 1134 times
Last edited by ricardo on Tue Aug 19, 2014 10:53 pm, edited 1 time in total.
User avatar
mediatechnology
Posts: 5444
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: 5532/4 THD

Post by mediatechnology »

Thanks for sharing this ricardo.
Sorry for the difficulty (now fixed) that prevented pdf files from being attached.

I can speak from experience that the performance of ICs are strongly process-dependent.
A close friend, who makes them, once likened it to baking a cake.

The Raytheon 5532 and 5534s were the worst.
I suspect that the line making them had gold contamination from equipment also used to process gold-doped logic parts.
They were noisy, gritty, distorted a lot and tended to overheat and go into thermal runaway.
I've replaced literally thousands of them in SSL consoles.

WRT the 18 MHz oscillation seen in the 5532/4562 test fixture in the GroupDIY link one of the first things I would look at is capacitive loading from the scope probe and a usual suspect supply bypass.

In many applications the LM4562 will be slightly noisier compared to a 5532 due to the input bias current cancellation circuitry thought to be used in the LMM4562/LME49710/LME49720/LME49860 etc.
The 1 kHz typical noise current for a 5532 is 0.7 nA/rootHz vs. 1.6 nA for the LM4562.
See this thread for information on the suspected bias current cancellation used in the LM4562 and LME49710-series: viewtopic.php?f=6&t=641
There's a copy of the NSC patent in the thread.
ricardo
Posts: 262
Joined: Sun Apr 10, 2011 7:24 am

Re: 5532/4 THD

Post by ricardo »

mediatechnology wrote:The Raytheon 5532 and 5534s were the worst.
I suspect that the line making them had gold contamination from equipment also used to process gold-doped logic parts.
They were noisy, gritty, distorted a lot and tended to overheat and go into thermal runaway.
I've replaced literally thousands of them in SSL consoles.
I now remember having this with another Raytheon "LN" device, OP27 IIRC.

It's a horrible moment when you realise you've just made several hundred defective products .. but you thank the stars they haven't been shipped. :shock:
User avatar
mediatechnology
Posts: 5444
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: NE5532 NE5534 THD

Post by mediatechnology »

It's a horrible moment when you realise you've just made several hundred defective products
I once had a "client" that got all of his SSL RC5534s replaced - thousands of them - at no charge for parts or labor over two years after the console was commissioned.
SSL picked up the tab for about 1800 Signetics NE5534s and paid me my hourly rate to replace the Raytheon RC5534s.

The accounting department, based out of their lawyer's office in Tulsa, later called to complain - literally chewing me out - about the $35 Fedex charge they had been billed for to have the parts shipped in.
I refused to ever go back there: The client was later investigated by the government and sued for fraud: http://en.wikipedia.org/wiki/Robert_Til ... nvolvement
Karma is a b^tch.
Fortunately the console turned up across the freeway and is enjoying life at MediaTech institute.
ricardo
Posts: 262
Joined: Sun Apr 10, 2011 7:24 am

Re: NE5532 NE5534 THD

Post by ricardo »

http://www.edn.com/design/audio-design/ ... ght-op-amp

This is the first bit of info that might make me want to use something else besides my beloved NOS Mullard 5532/4s carved from Unobtainium by Southampton virgins for general purpose.

Circa 1980, I did an internal Calrec report on TL07x / NE5532/4, mainly to do with non-inverting amps and the THD due to CM. At the time, the practical conclusion was simply not to use TL07x non-inverting. 5532/4 would just start showing THD rise at HF with follower or non-inverting 3x gain 600R load IIRC .. but with the gear we had at the time, was practically blameless inverting.

Self shows LM4562 is very resistant to CM THD.

He also correctly identifies (on the next page) RIAA preamps as the application that most shows off the extra noise with EVIL Ib cancelling.

Wayne looks at other caveats with Ib cancelling in 4562 viewtopic.php?f=6&t=641.

Dunno if 4562 has the Ib cancelling noise problem. It's possibly better cos easier today to make LN current sources ... but I might be pontificating from the wrong orifice.

I would take Doug's claims that 4562 is less fussy about decoupling than 5532/4 with a HUGE pinch of salt.
User avatar
mediatechnology
Posts: 5444
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: NE5532 NE5534 THD

Post by mediatechnology »

Thanks for the link ricardo.

Having used them I really like the LM4562, LME49710, LME49720, LME49860.
I've never used the quad version, the LME49740.
Given that the NE5532 is rapidly disappearing in through hole DIP the LME49710-series is becoming more attractive.

About the only limitation I see, compared to the NE5532, is the +/-17V (34V) supply limitation.
For +/-15V rails they're fine.
From having attempted to run an LME49720 on 36V I can assure you that they really mean it when they spec 34V as the maximum.
What you get in return with the LME4720 vs the NE5532 is the ability to operate on supplies as low as +/- 2.5V.

The LME49860 (dual) however can operate from 36V supplies with a 44V maximum.
I've never had any issues running the LME49860 on 36V.
User avatar
JR.
Posts: 3700
Joined: Sat May 24, 2008 7:21 pm
Contact:

Re: NE5532 NE5534 THD

Post by JR. »

I have surely shared this before, (I think I've seen this specific discussion elsewhere on the WWW), but +1 to Wayne's comment about process dependency of marginal (noise, gain. etc) performance. Back while I was working at Peavey we had a house number dual op amp in the system, that was a selected 5532. I even remember the house part number 70403678, affectionately known inside as 3678.

One odd thing about the selection process was a relative noise level test, not absolute. I do not recall the exact test but something like if the 10Hz noise was more than x dB higher than 1kHz noise the part was rejected. I don't recall who the vendor selling these was, it might have been ***Raytheon, it wasn't Signetics or a primary manufacturer, as I recall.

The rationale for this controlled LF noise spectral envelope, was to manage how a mixer would "sound" with some bus amp turned WFO. With significant 1/F noise the noise floor sounds very unhealthy, like a circuit "fixin to blow up". While the more well behaved noise floor dominated by higher frequency noise sounds more benign. A positive side effect of selecting out devices with high 1/F noise we probably dodged a few other subtle flaws and/or reliability issues.

JR

***I do not know the history on this selected part but is seems logical that Peavey might have negotiated the test after having a problem with Raytheon's part. More logical would be to just black-list the Raytheon part, but maybe they came back with a better part, for a competitive price (just guessing). As I recall when I was using them they were slightly more expensive than stock 5532s, because I ended up bringing in the standard part too, and using them for low gain sockets to save a few cents per. FWIW during this time there was a general clean-up going on in the semiconductor industry wrt process so the selected parts may have been useful for large IC users back in the day for reliability. Imagine a large mixer production line using thousands of these per shift, it wouldn't make many DOAs to be noticed.
Cancel the "cancel culture", do not support mob hatred.
User avatar
mediatechnology
Posts: 5444
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: NE5532 NE5534 THD

Post by mediatechnology »

Also found this some time ago in Self's book regarding 5532 follower distortion being slightly lower with a 1K series resistor.
Apparently there's a source impedance "sweet spot."

Image
Douglas Self, "Small Signal Design," NE5532 Follower CM Distortion vs Source Resistance
User avatar
JR.
Posts: 3700
Joined: Sat May 24, 2008 7:21 pm
Contact:

Re: NE5532 NE5534 THD

Post by JR. »

I believe in pushing the envelope but probably would have stopped at 0.00065% THD. 8-)

One of my complaints about the uber low THD new opamps is that the only way they can measure them is at high gain, so not a real test of linearity in follower and low gain, higher source impedance applications.

I have had some discussions about this with Sam and IIRC he has some plots of linearity wrt source impedance in his data base.

JR
Cancel the "cancel culture", do not support mob hatred.
ricardo
Posts: 262
Joined: Sun Apr 10, 2011 7:24 am

Re: NE5532 NE5534 THD

Post by ricardo »

The reason why this Rs dependence of THD doesn't occur with shunt feedback (inverting) is cos the main cause of CM THD is wonky Cob in the input transistors. In series feedback (non-inverting) amps, Rs reacts with this wonky Cob .. especially if Rs is mismatched on the 2 inputs. Cascode i/ps (and shunt feedback) minimize changing voltages across the wonky Cob.

A very good demo of this is input stages in simple power amps. Small NPNs usually have smaller & hence less wonky Cob than their PNP complements.

Just swapping a PA design so it uses NPN inputs usually reduces THD by a significant amount for free .. you don't have to use EVIL expensive input cascodes :D

The main THD reduction is actually cos the PSR improves so the yucky Class B and PSU stuff doesn't get back into the LTP inputs. The other stuff in your simple power amp has to be good though to see this.

Need to say NPN's better LF hoe is also partly responsible but wonky Cob is the main mechanism for this.
Post Reply