DIYstompboxes.com

DIY Stompboxes => Building your own stompbox => Topic started by: seedlings on March 01, 2012, 01:09:10 PM

Title: LM308N vs...
Post by: seedlings on March 01, 2012, 01:09:10 PM
I've searched and get bits here and bits there, but my eyes are crossed.

Can you give me an experienced comparison between a LM308N and some other opamps?  I have TL072, LM833N on hand.  I've read that a TL071 are an OK sub for the LM308N, but I've also read some swear by 'the sound' of the 308.

Is there a reference file/post of how common (and uncommon) opamps compare in audio circuits?

CHAD
Title: Re: LM308N vs...
Post by: Govmnt_Lacky on March 01, 2012, 01:17:41 PM
If I remember correctly, the thing that makes the LM308N so special (especially in RAT-type circuits) has to do with the slew rate of the op amp.
Title: Re: LM308N vs...
Post by: seedlings on March 01, 2012, 01:21:07 PM
If I remember correctly, the thing that makes the LM308N so special (especially in RAT-type circuits) has to do with the slew rate of the op amp.

I read that.  The LM833N says Very High Slew Rat 5V/us min, 7V/us typ.  But it has to be different somehow (besides being a dual opamp) or the LM308N wouldn't cost 9x more.

CHAD
Title: Re: LM308N vs...
Post by: electrosonic on March 01, 2012, 01:41:53 PM
I believe the LM308 is an older op amp with worse specs than the ones you mention. Works great for a distortion pedal, so-so for a hi-fi. It is more expensive because it is obsolete and supplies are finite.

Andrew
Title: Re: LM308N vs...
Post by: zombiwoof on March 01, 2012, 01:44:06 PM
If I remember correctly, the thing that makes the LM308N so special (especially in RAT-type circuits) has to do with the slew rate of the op amp.

I read that.  The LM833N says Very High Slew Rat 5V/us min, 7V/us typ.  But it has to be different somehow (besides being a dual opamp) or the LM308N wouldn't cost 9x more.

CHAD

I'm sure the price difference has to do with the limited availability of the LM308's, which is the reason they changed the RAT opamp to the OP07 (which reportedly doesn't sound as good, I don't know because I only have LM308 RATs).  This, and the fact that RAT users in general are demanding the LM308 for their RAT pedals.  In other words, it's a supply/demand issue.

Al

P.S. posted this at the same time as the post right above mine.
Title: Re: LM308N vs...
Post by: newfish on March 01, 2012, 02:22:13 PM
Having auditioned various Op-amps in a 'Distortion +' style circuit, I can say the 308 and 071 are chalk and cheese.

The TL071 gives more top-end (FET-based - higher input impedance?), and is cleaner than the 308 - even at '10'. 

The 308 is *very much* all about the grind.

Oddly, the humble 741 sits comfortably between these extremes.

Hope this helps!
Title: Re: LM308N vs...
Post by: seedlings on March 01, 2012, 02:23:59 PM
Having auditioned various Op-amps in a 'Distortion +' style circuit, I can say the 308 and 071 are chalk and cheese.

The TL071 gives more top-end (FET-based - higher input impedance?), and is cleaner than the 308 - even at '10'. 

The 308 is *very much* all about the grind.

Oddly, the humble 741 sits comfortably between these extremes.

Hope this helps!

Hmm... I'll order a couple to see.  Hate to pay $7.xx per chip with shipping... oh well.

CHAD
Title: Re: LM308N vs...
Post by: Mark Hammer on March 01, 2012, 02:50:53 PM
The basis for use of the 308 was made clear in an excerpt from Art Thompson's "Stompbox" book reprinted in Guitar Player some years back (http://www.amazon.com/Stompbox-History-Guitar-Flangers-Phasers/dp/0879304790#_ ).  They reprinted the interviews with several pedal designers, including the folks at Proco   Part of the origins of the Rat was essentially a component-value error.  The designer thought h was sticking in a larger value ground-leg resistor, but inadvertently used a low-value one and set the gai skyrocketing.  Most importantly, the gain it was set to vastly exceeded the gain-bandwidth product of the chip, and this interesting sort of undertone was produced.

The Rat's tone...at least as much of it comes from the choice of chip...is a function of a) the slew rate, b) the compensation cap value, c) the gain-bandwidth product of the chip, relative to the gain setting.  Keep in mind that all chips have a slew rate, and that their capacity topass AC with greater amplitude at the output than at the input (i.e., gain) decreases as frequency goes up.

Look at the graph in the upper right corner on page 14 here: www.national.com/ds/LM/LM108.pdf  where you'll see the open-loop gain - the gain-bandwidth product.  With a 30pf compensation cap, you'll see that the chip is simply not capable of producing the gain of over 3000 for content above 1.5khz when the stock Gain control is maxed.  Where the LM308 can only produce a max of 50db gain at 1khz, the CA3140 can do 70db, the NE5532 looks like it can do around 78db,  the LF441 will do 60, and the NJM4558 will do 70.  The datasheet for the OP-07 shows it having an open-loop gain-bandwidth product of 60db at 1khz.  maximum gain is obviously lower for all of these at higher frequencies.

So part of the sound of a Rat is essentially the chip collapsing under pressure, and the sound of that as clipped by diodes.

Certainly, one can get distortion by using ANY op-amp to amplify the signal several hundred-fold and passing that by some diodes.  One can also get an interesting quality or tone of distortion by using any old op-amp and the dual ground-leg arrangement of the Rat to produce additional boost for upper mids and beyond.  But the true sound of a Rat requires that the chip be as limited in capacity as the 308.  There may well be others with that same sort of limitation, but I don't know what they are
Title: Re: LM308N vs...
Post by: seedlings on March 01, 2012, 03:14:23 PM
^ Thank you Mark.  I seem to get a grasp of this stuff for fleeting moments.  I'l re-read your post a bazillion times.

CHAD
Title: Re: LM308N vs...
Post by: blackcorvo on March 01, 2012, 03:59:40 PM
Ever tried the TL070 ?
Title: Re: LM308N vs...
Post by: Mark Hammer on March 01, 2012, 04:38:08 PM
because the TL070 uses an externally compensation cap, it might be possible to mimic the performance of the 308, but the datasheet doesn't really provide enough information to know what cap value to use to do that.
Title: Re: LM308N vs...
Post by: ashcat_lt on March 01, 2012, 05:12:16 PM
I followed a suggestion from some old post on this board and used a TL080 with a 180pf(?) comp cap.  Can't say that it sounds "right" but I'm happy with the results.
Title: Re: LM308N vs...
Post by: Mark Hammer on March 01, 2012, 05:24:05 PM
I followed a suggestion from some old post on this board and used a TL080 with a 180pf(?) comp cap.  Can't say that it sounds "right" but I'm happy with the results.
1) Happy with the results is the primary target.  Hell, you may not even like an "authentic" Rat.

2) There may well be other chips that do it.  We know the comp-cap value for the Rat, using a 308.  We don't know the precise value for other chips.  However, 180pf may well just be the functionally comparable value.
Title: Re: LM308N vs...
Post by: ashcat_lt on March 01, 2012, 07:50:52 PM
Well, I did like the Rat I bought in 94.  Never had any need or desire to know which opamp was in it at the time.  That one was lost or stolen or something so I cant do a side-by-side comparison.

I'm happy with the one I built with the 080.  Course I was also happy with the very similar one I built with an lm324...  What I don't like is the fact that one diodes lifted (for extremely assymmetrical clipping) sounds and looks exactly like when both diodes are lifted.  There was a thread around that said something about the coupling caps floating up, which sort of led to my question on this thread. (http://www.diystompboxes.com/smfforum/index.php?topic=96354.0)
Title: Re: LM308N vs...
Post by: seedlings on March 01, 2012, 11:07:03 PM
So I just breadboarded the rat with the LM833N with 2N5457 fets for the buffers.  I've never played through a Rat of any sort, but this sounds darn good.  First used 1n4148s, then 1N60P schottkys, red LEDs, and asymmetrical combinations.  The 1N60Ps were super compressed, and I liked those second to... 1N4007s.

Anyway, I had no idea the Rat had so much low end fuzzy growl.  Seems like the sound clips I hear are thin or tinny- sort of like a DS1.  I thought I still had some TL072s, but can't find them to try.  Waiting on the LM308Ns and more TL072s for comparison.

CHAD
Title: Re: LM308N vs...
Post by: PRR on March 02, 2012, 06:41:48 PM
> LM108 ... open-loop gain - the gain-bandwidth product.  ...the chip is simply not capable of producing the gain of over 3000 for content above 1.5khz when the stock Gain control is maxed

I did this sim of a '308-like chip with several Rat-like gain settings:

(http://i.imgur.com/W5vD9.gif)

As you say, as we see, "huge boost" just does not happen at higher frequencies.

Use a "better" op-amp, it does; and burns your ears.

There is probably also something happening beyond simple linear frequency response when a '308 is slammed. Clipping/slewing may be neat, clean, symmetrical, or lopsided or lumpy.

If '308 becomes impossible, try LM301 with 100pFd to 300pFd compensation cap.
Title: Re: LM308N vs...
Post by: davent on March 02, 2012, 07:32:47 PM
I believe these are metal can type packages. Have no idea about this company's minimum & shipping charges, shows six in stock.

http://rcfreelance.com/searchresults.php?email=websales@rcfreelance.com&partnumber=LM308AH
Title: Re: LM308N vs...
Post by: davent on March 02, 2012, 07:46:03 PM
I believe these are metal can type packages. Have no idea about this company's minimum & shipping charges, shows six in stock.

http://rcfreelance.com/searchresults.php?email=websales@rcfreelance.com&partnumber=LM308AH

If you're ever in Toronto, Honson Computers used to have a supply of LM308's but... can't say whether that's still the case.

dave
Title: Re: LM308N vs...
Post by: greaser_au on March 03, 2012, 09:12:22 AM
I used to have easy access to a lot of stuff like this at work, but alas, no more - we no longer do component level repair work & we sold off/trashed all of our stock a few years back... :(   
However, I do note that you can buy a handful of LM308Ns for under a dollar each (or under $2 in singles)  shipped from Asia on e-bay...

best wishes,
david
Title: Re: LM308N vs...
Post by: Mike Burgundy on March 03, 2012, 10:27:33 AM
Just a thought - if the slewrate is the big thing, an LM324 might be worth a look. A 301 is closer to an 071 than a 308. Hang on, a 741 is also slowish, and not popular for a RAT if I recall correctly. Thus proving there's more than just the slewrate....
Title: Re: LM308N vs...
Post by: DavenPaget on March 03, 2012, 11:02:02 AM
Just a thought - if the slewrate is the big thing, an LM324 might be worth a look. A 301 is closer to an 071 than a 308. Hang on, a 741 is also slowish, and not popular for a RAT if I recall correctly. Thus proving there's more than just the slewrate....
Why ? Because the slewrate can be adjustable to liking . The limited slewrate by the common 30pf cap causes a ghostly sound in the background , that's what the RAT is famous for .
Title: Re: LM308N vs...
Post by: seedlings on March 03, 2012, 11:33:37 AM
What's a 'famous song' featuring the ProCo Rat?

CHAD
Title: Re: LM308N vs...
Post by: DavenPaget on March 03, 2012, 11:46:54 AM
What's a 'famous song' featuring the ProCo Rat?

CHAD
Don't know .
Title: Re: LM308N vs...
Post by: ashcat_lt on March 03, 2012, 03:59:45 PM
Holy parallel threads! 

The lm324 will swing about 1.2 volts further in the negative direction than positive.  The clipping from the opamp will be asymmetrical.  I don't know if this is a noticeable difference, really.  The two rat based circuits I've built around lm324 sound pretty cool.  Don't know about any "ghost sound", though.
Title: Re: LM308N vs...
Post by: Mark Hammer on March 03, 2012, 05:11:11 PM
As a matter of fact, if you look at the Boscorelli book, the Distort-o-matic IV exploits that property of the 324 by deliberately cascading four sections of it as a clipping element.  How on earth four unity-gain inverting op-amps give you clipping is beyond me, but then Boscorelli notes that it won't be an especially strong effect. 
Title: Re: LM308N vs...
Post by: StephenGiles on March 04, 2012, 01:02:15 PM
Just a thought - if the slewrate is the big thing, an LM324 might be worth a look. A 301 is closer to an 071 than a 308. Hang on, a 741 is also slowish, and not popular for a RAT if I recall correctly. Thus proving there's more than just the slewrate....

I had a spare 324 amp when building a VCF a few years back, so I used it for a Rat which sounded almost identical to my 308 version.
Title: Re: LM308N vs...
Post by: StampWah on April 09, 2018, 01:14:12 AM
Hi All ! I'm looking for dual version of 308. Any thoughs?
Title: Re: LM308N vs...
Post by: PRR on April 09, 2018, 09:11:59 PM
Welcome.

There never was a dual LM308A.
Title: Re: LM308N vs...
Post by: Rob Strand on April 09, 2018, 10:12:08 PM
This idea isn't so well known.


(https://s14.postimg.cc/3m6joau25/Shaping_Clipped_Opamp_Harmonics_v1.png) (https://postimg.cc/image/3m6joau25/)
Title: Re: LM308N vs...
Post by: Rob Strand on April 09, 2018, 11:00:42 PM
Quote
324
358 is the dual version.
Title: Re: LM308N vs...
Post by: amptramp on April 10, 2018, 07:00:10 PM
All of us rejoiced when the LM308 came out.  Now we had an op amp with less than 0.5 mV input offset voltage and low drift as well.  We could do DC-coupled circuitry like absolute value circuits without needing offset adjustments and that was a good thing because offset adjustment is possible but not really catered for.  But not much of this is important in stompbox circuitry where AC-coupling is the norm.  But there were a few things you had to take into account in a design:

1. There were back-to-back diodes between the inputs so a circuit that pulled the inputs to different voltages had to have a resistor on the input to avoid burning these diodes (really, diode-connected transistors).  This is necessary because the input transistors are superbeta types with a very thin base.  The hfe might be around 1000.

2. Input common mode range was within one volt of the rail in either direction.  Not quite rail-to-rail but close.

3. The output would swing to within 1 volt of the rail with a load of 1.4 mA.

4. The supply current was 0.3 mA to a max of 0.8 mA.  Not bad and better than a lot of op amps.

5. Bias current would usually be 1.5 nA.  More than FET amplifiers but much less than other contemporary bipolars.

6. Don't skimp on the compensation.  For audio, feedforward compensation is better because it gives you more bandwidth but the standard compensation (both versions) works and keeps the bandwidth down if that is what you want.  I once had a circuit that required 100 pF compensation to ground and it still oscillated.  The problem?  The capacitor actually measured 91 pF.  A true 100 pF and above fixed it.

7. If you rely on an uncharacterized property (like gain-bandwidth), you will get different results for every device which goes against why you use op amps in the first place.
Title: Re: LM308N vs...
Post by: PRR on April 10, 2018, 10:34:43 PM
LM308 had lower bias current than contemporary JFET inputs at high temperature. Some folks need it hot.

The lingering 'feature' of the LM308 is its poor speed which softens audio clipping. There's a billion other ways to do this. But its use in some Classic Pedals ensures a market long after it could have been forgotten.
Title: Re: LM308N vs...
Post by: rankot on April 11, 2018, 03:05:51 PM
And what is real slew rate of LM308? I have two datasheets and none of them mention it.

Looking at open loop freq. response for LM301, LM308 and LM324 gives almost the same result (55dB@1k) with 30pF compensation cap.

UA776 could be a good candidate (slew rate is 0.8V/ms), but datasheed doesn't contain open-loop freq. resp.
Title: Re: LM308N vs...
Post by: Rob Strand on April 11, 2018, 07:13:47 PM
Quote
And what is real slew rate of LM308? I have two datasheets and none of them mention it.
The value in my head is 0.3V/us [Edit: maybe 0.03Vus] maybe for C=100pF?.   However it depends on the compensation cap.  There a lot of info scattered in the National Semiconductor Linear Applications Handbook.   You can derive it from the Large Signal Response graph  SR  = 2* pi * f_max * Vpk_max /1e6 [V/uS]; use Vpk not Vpk-pk.

One of the reasons for having uncompensated opamps is you can tweak the compensation cap to get more speed.   If you have a high gain circuit you don't need such a large compensation cap for stability (against feedback loop oscillation).  So that means you can get a higher gain-bandwidth and higher slew-rate.    Many designs just use the unity gain value.    IIRC, Philips/NXP NE5534 datasheet and associated applications notes had some nice info about that.

----------------
LM108/LM108A Slew-rate calculation from "Large Signal Response" graph

For C = 30pF,
Vpk_max = 14.3Vpk,
f_max   = 1.4kHz
SR = 2 * pi * Vpk_max * f_max / 1e3 = 2 * pi * 14.3 * 1.4e3 / 1e6 = 0.13 V /uS

For C = 100pF estimate SR = (30p/100p) * SR(30pF) = (30/100) * 0.13 = 0.038 V/uS


Title: Re: LM308N vs...
Post by: PRR on April 11, 2018, 11:10:27 PM
> And what is real slew rate of LM308? I have two datasheets and none of them mention it.

Read the pictures. "Voltage Follower Pulse Response".
Title: Re: LM308N vs...
Post by: Rob Strand on April 12, 2018, 12:05:33 AM
Quote
Read the pictures. "Voltage Follower Pulse Response".

Actually that method will match the SR value quoted in most datasheets.