LM308 query

[Digital Larry]

I have come across a fairly unique mod to the RAT circuit.  This one appears to use the H metal can package but any markings were sanded off.

For sure I can see that the RAT topology is going to have a dependency on the op-amp's performance simply because it has a ton of gain and it slams the signal into the rails which a Tube Screamer does not.  The behavior of the circuit under these conditions is of course not well documented, because any "sane" designer would not design a circuit that did such things.  We have the slew rate and freq response which are set by the compensation cap.

I read some threads here where folks compared the LM308 to a TL070 or an OP07 or a 5532 family part but there's little in the way of objective measurements, such as what the output looks like when you slam this signal into the rails.  Comparisons of pedals on YouTube are similarly unscientific, what I've found anyway, as it's not clear that the ONLY difference between one pedal or another is the op-amp.

Based on following a few threads here I'm not convinced that the LM308 has any inherent mojo, but I also like to be open minded, so this is where you can tell me that it's the only way to go, and if so, why that is.

Let's look at the datasheet from National:

http://pdf.datasheetcatalog.com/datasheet/nationalsemiconductor/DS007758.PDF

Page 4, "Open Loop Frequency Response".    From this you can get the freq response for lower gains using feedback which the Rat does.

For cf = 30 pF, at 40 dB of closed loop gain the bandwidth is about 2 kHz.  Slightly more gain, slightly less bandwidth and vice versa.  So it may be that the LM308 compensation circuit happens to give a nice auto-adjustment of frequency response in the audio range as you crank the gain.

[pinkjimiphoton]

try a metal can lm301 in a rat. that's what i have been using , they sound great and i suspect may have been some of the chips used with the markings sanded off way back in the early days of the design.

i can't discern an audible difference, but YMMV. i've a/b'd it with originals and can't tell the diff.

[ashcat_lt]

When I was coding my Rat sim, I had assumed that GBW was just a function of slew rate limiting, but it didn't sound right until I also added a low pass for the GBW.  I think I decided this had to come before slew limiting, but I'm not sure that I was correct in that.  Does anybody know this?  I also wonder if slew rate limiting is curvy.

Anyway, you definitely can tweak the frequency response with that cap.  I personally don't know what else that cap does.  I tend to think it can affect stability or noise response of something else fairly important, but I don't know enough about it to say what happens or how much wiggle room you've got.  I can't imagine you'd blow anything up trying some different values.

Honestly, I'd like to know more about how this works.  It's awful complex, and while the straight LPF that I used seems to work - even in tests against its hardware counterpart - I wonder how the filters in the feedback loop interact with this. 

[Digital Larry]

Thanks for the input.  Glad to know there's an externally compensated op-amp still in production.  SMT too!

[Digital Larry]

When I was coding my Rat sim, I had assumed that GBW was just a function of slew rate limiting, but it didn't sound right until I also added a low pass for the GBW.  I think I decided this had to come before slew limiting, but I'm not sure that I was correct in that.  Does anybody know this?

College was a LONG time ago, but... I think you'd put slew rate limiting at the end.  Your GBW curve simply shows the frequency response you could get with different gains.  Slew rate limiting will happen at a high enough frequency and high enough amplitude.  Theoretically you can eliminate slewing at low frequencies by bringing down the output amplitude.

[ashcat_lt]

Thanks for the input.  Glad to know there's an externally compensated op-amp still in production.  SMT too!

Oh yeah, PS, I took a hint from some long buried thread and used a TL080 with a much bigger comp cap in my meatspace rat thing.

[ashcat_lt]

College was a LONG time ago, but... I think you'd put slew rate limiting at the end.

I agree with that.  No matter what the input x gain turns out to be, the thing can only change so quickly.

What trips me up is that the Rat circuit "asks for" different gain at different frequencies, so, like, WTF?

[thermionix]

try a metal can lm301 in a rat. that's what i have been using , they sound great and i suspect may have been some of the chips used with the markings sanded off way back in the early days of the design.

Why wouldn't they be metal can 308s?  I've built 2 Rats so far, both with NOS Motorola metal can LM308AHs.  They both sound great, but I don't have anything to compare them to, and it's been about 30 years since I last played a real ProCo Rat.  I now have 4 new production metal can 308s that I got from Ebay...I hope they're real, lol.

What else is built (or can be built) with LM308s?  Anything really simple I can throw on the breadboard to test these new cans out?  I have another Madbean Runt board, but don't feel like building another Rat until I get commisioned to do so.

[PRR]

> slew rate ....  GBW

Related but not the same.

There is a compensation cap.

In the usual scheme:

Input stage current limits how fast the cap can charge. Slew-rate.

Input emitter resistance against the cap sets a frequency response. GBW.

In "naked emitter" input stages, current and resistance are linked; but when you compensate to avoid the stray poles, the slew-rate sucks. Most input stages have added resistance (or stacked junctions) to offset the natural linkage.

JFET input stages naturally have "extra emitter resistance" and often hit a fine balance of GBW and slew without much fuss.

> 40 dB of closed loop gain the bandwidth is about 2 kHz.

Good point.

Any externally compensated opamp can be over-compensated to be as slow as a '308. For the usual suspects, 300pFd is a starting point to slug it down. Slew will drop, but the '308 is a slew-slug, so this is going the right way.

[Digital Larry]

> slew rate ....  GBW
Input stage current limits how fast the cap can charge. Slew-rate.

Input emitter resistance against the cap sets a frequency response. GBW.

Slew-rate limiting will create non-linear distortion artifacts.  High frequency roll off will not.

[ashcat_lt]

Slew-rate limiting will create non-linear distortion artifacts.  High frequency roll off will not.

Yes.  For large signals, slew rate acts like a low pass also.  Again I ask if this is curvy or just a ceiling...

But again I'd ask: the Rat has a feedback structure that asks for a lot of gain in the upper midrange, a lot less in the low end, and a rolloff in the treble region.  So how does GBW touch that?  Like, is there more bandwidth at 12K or 100H than at 3K?  But what the hell does that mean?!? 

[Digital Larry]

Look for "Voltage Follower Pulse Response" in the data sheet to see what slew rate limiting looks like.

Regarding your other questions about frequency shaping, it's starting to get into more advanced and less intuitive areas.  I mostly remember the rules of thumb and engineering approximations rather than the deep theory.

http://www.radio-electronics.com/info/circuits/opamp_non_inverting/op_amp_non-inverting.php

This gives some basic ideas.  Keep in mind that R1 and R2 can be in general substituted by "impedances" which could be a cap across a resistor, etc.  This is how the frequency shaping is accomplished and it presumes that you have open-loop gain to burn at the frequency (as the extra gain is what makes feedback work). 

[PRR]

> High frequency roll off will not.

When you run out of GBW, you lose the distortion-reduction effect of NFB. The open-loop THD of chips may be poor or unbelievably poor. It should be audible.

A point: over-compensation will keep the amp semi-clean even when the external NFB has pooped-out. The compensation _is_ a NFB, which keeps the VAS stage fairly linear at higher F. The output stage can crap-up after this. Too much to think about.

[digi2t]

Whilst researching the LM308, I happened upon this gentleman. Wow... as they say, line between genius and insanity is very fine indeed!

https://en.wikipedia.org/wiki/Bob_Widlar

[ashcat_lt]

When you run out of GBW, you lose the distortion-reduction effect of NFB. The open-loop THD of chips may be poor or unbelievably poor. It should be audible.

I don't really know what to do with this.  This is more distortion on top of slew rate distortion?  I tend to think that the lowpass action naturally reduces the harmonics produced, but maybe not.  Ultimately, in a rat, we've got plenty distortion that a little more won't barely be noticed.

The compensation _is_ a NFB...

Can we just think of this cap as more or less parallel to the external NFB?  That would make my life a lot easier. 

[bloxstompboxes]

Whilst researching the LM308, I happened upon this gentleman. Wow... as they say, line between genius and insanity is very fine indeed!

https://en.wikipedia.org/wiki/Bob_Widlar

Why has there been no movie made of this person? I mean the goat, or rather, the sheep, come on. lol.