MUFF WIGGLER Forum Index
 FAQ & Terms of UseFAQ & Terms Of Use   Wiggler RadioMW Radio   Muff Wiggler TwitterTwitter   Support the site @ PatreonPatreon 
 SearchSearch   RegisterSign up   Log inLog in 
WIGGLING 'LITE' IN GUEST MODE

Protection resistor placement?
MUFF WIGGLER Forum Index -> Music Tech DIY  
Author Protection resistor placement?
Navs
What's the best place for an output protection resistor - inside or outside of the output buffer's feedback loop?

I'm building Ken Stone's CGS18 Drum Simulator on stripboard, minus the harmonics section. On the schematic, Ken has a 1K resistor after the feedback point.

http://cgs.synth.net/modules/cgs18_drum.html

In the schem to his DC Mixer, he has it inside the loop:

"The 1k (330 ohms on later boards) at the output simply offers some protection to the op-amp in the event of a short circuit, which happens frequently when patch cords are being used. It is inside the feedback loop, so adding load will not affect the output, so it can quite happily drive a multiple jack."

http://cgs.synth.net/modules/cgs04_mix.html

Is there a reason why it's outside the loop on the drum sim?
wsy
If the resistor is outside the feedback loop, then any load from outside (i.e. what you patched to) limits current according to Ohms Law. If you're running
Dotcom voltages, (+/- 15V) then 1 Kohm limits current to at most 66 mA,which is more than most op-amps are specced to allow (nominal on a
TL07x opamp is 40 mA before it self-limits), so the op-amp still goes into current limiting internally - and that's what happens every time you plug in any
two-conductor cable (i.e. 1/4" or 1/4". Bananas are the only cables that don't short during plug-in).

[[[ EDIT: next poster is correct: the current is 15 mA, not 66 mA. I must have been sober or something equally bad. So, no clipping, but if you have
a warm opamp, it might go to clipping anyway. But it's still not horrible. ]]]


If the resistor is inside the feedback loop, the max current is still limited by the op-amp, but until that happens, the output voltage of the op-amp servoes to
keep the voltage after the resistor correct.

So, for shorting, either works about as well as the other.

However, if you plug two outputs together accidentally, then with the resistor after the feedback loop, it'll act like a mixer set 1:1 (well, up until one of the
op-amps goes into clipping); at least a musical effect, and no sparks should result.

With the resistors inside the feedback loops, the two op-amps will be in some sort of crazy cage-match, fighting each other, and could conceivably
go into oscillation in the ultrasonics (all depending on the rest of the circuit topology) and bizarreness may happen - especially if your downstream speakers
don't have good ultrasonic filtering, as you may then blow the tweeters.

At least that's my guess- and it's only a guess. But I put my protection resistors outside of the feedback loops.

- Bill
daverj
^
|
| what he said.

Well, except he calculated the current wrong. At +/-15v signals, a 1k outside the feedback loop will limit the current to +/-15ma if shorted to ground or 30ma if shorted to a power rail. When connected to another module with a similar 1Kseries resistor it's limited to +/-15ma max. So the op amp shouldn't go into clipping.

With it outside the feedback loop there is always some amount of attenuation to the signal. The more the load, the more attenuation. But there should never be any clipping in the amp, no matter the load.

When inside the feedback loop the feedback tries to maintain the output voltage, so the output voltage does not change with load, until the load is large enough that the amp reaches it's clipping point. Most modules have a very small load, so you would have to split to hundreds of modules before you would see clipping.

And inside the loop, connecting two module outputs together is exactly as he described. Some weird battle between the amps, both trying to correct for the others voltages. It would not be pretty.

Also be aware that modern amps, like the TL072 are internally protected from short circuit, so neither method will harm them. But some older op amps (still used in some old DIY designs that people occasionally build) can't handle those higher output currents. So it's not a good idea to change the design of an old design from having a resistor outside the loop to one inside the loop if there are older non-protected op amps used.
Navs
Thanks for the comprehensive answers, Bill & Dave!

I'm using an 074, but will keep the 1k outside the loop.
Dr. Sketch-n-Etch
OK, I'll weigh in here as well...

Most opamps don't really need short circuit protection (as clearly stated in their datasheets -- usually in the "Absolute Maximum Ratings" section). It will typically say that the allowable short circuit time is unlimited. However, one does have to be careful that the IC doesn't overheat. I still think it's a good idea to use current limiting resistors. In any case, I believe that TL07X tops out at about 26mA even without current limiting.

For audio outputs, putting 1k after the feedback junction is good. For CV outputs, a resistor here will act as a divider with the input resistor on the receiving end and create a CV error (important if pitch is being controlled). In those cases the resistor should go inside the feedback junction. It is also easier to route inside resistors on PCBs.

However, the main difference is in how the opamps respond to downstream capacitive loading. An "outie" resistor provides very good compensation for capacitive loading (and a long patch cable is often enough downstream capacitance to create oscillation problems). And "innie" resistor does nothing for capacitive loading, and in fact can even make matters worse. This is why I apply a special capacitive loading compensation network on all my opamps when I'm serious about not creating CV errors. Basically, my preferred circuits are as shown in the following diagram (note: the 220R "outie" resistors can be 1k):



Hope this helps.
wsy
Yes, using a capacitor from Vout to Vin- is a good trick if you are worried about going into oscillation, but TL07x opamps usually aren't fast enough to cause a
problem unless there's weirdness downstream. I've never needed it with TL07x chips as long as I had a good power supply (and perhaps a bypass cap over the opamp)

The capacitor from Vout to Vin- will slow the response down and hopefully prevent oscillation. Hopefully. :-) It *does* work; I did exactly that on a project at
work where the upstream was a relatively fast (15 MHz, OP353?) rail-to-rail opamp, but the downstream was a pair of power FETs that had gain up to 50 or so MHz.
As you might expect, there wasn't enough phase margin in the system and it would go into oscillation pretty easily without the capacitor. Yes, I am
oversimplifying, but that's the general circuit layout at least.

As to tuning issues: why are you worried? The first stage of the typical VCO is going to either be a summing junction to a zero-volt potential summing
node ( in which case your tuning trimmer will flatten response to 1 V/octave), or it will be a unity gain follower with FET inputs (maybe with protection diodes)
in which case any sub-megohm output impedance won't make any difference.

On the other other hand, Dr. E-a-S has built more synth modules than I have, so maybe I should STFU and listen.

- Bill
Dr. Sketch-n-Etch
Well, wsy, I typically don't have too much trouble with oscillations using TL07X, but it is definitely a problem in SMD layouts, particularly if one tends toward larger resistors in the feedback loop to avoid drawing too much current. This I've learned the hard way, by bitter experience. Indeed, there are many places in SMD circuits where I'm finding that I need to add feedback capacitors where I would never have to worry about it in my through-hole DIY layouts.

As far as the other stuff, it's all based pretty much on experience. Opamps can do some pretty crazy things, and not all of them are written about in textbooks. It sometimes takes a pretty keen eye and a fair bit of thought to figure out why a certain circuit is giving poor results. However, I have learned that small feedback capacitors are probably the single most important weapon in my arsenal for making circuits behave themselves.

(Now, having said that, I must confess that I really don't understand why this is necessary with TL07X, since it is already internally compensated. I'm not smart enough to understand all the subtleties around this issue, but I do at least know what to do about it when things aren't working properly.)

Concerning CV errors and tuning, yes, most CV circuits in VCOs are based around FET input opamps. However, they are generally organized as inverters, so the input resistor becomes the defacto input impedance. If this resistor is 100k, and the CV source has a 1k output resistor, then there will be roughly a 1% error in the CV, and therefore in the tuning. Using an "innie" resistor eliminates this error.
gwaidan
Dr. Sketch-n-Etch wrote:
However, the main difference is in how the opamps respond to downstream capacitive loading. An "outie" resistor provides very good compensation for capacitive loading (and a long patch cable is often enough downstream capacitance to create oscillation problems). And "innie" resistor does nothing for capacitive loading, and in fact can even make matters worse.


This x 10. Example-I repurposed the (TL074-based) CV mixer circuits on my TW LPGs as output mixers for a quad build and noticed there was what sounded like intermodulation distortion on the outputs when putting them into my trusty Peavey guitar amp. Multed the outputs into a CRO and there was a clear ultrasonic waveform superimposed on the outs. Hacked the circuit to pull the 1k resistor outside of the feedback loop and everything was peachy.

The TL07x/08x series of chips are only just internally compensated for unity gain-throwing an extra compensation cap into the loop like Dr Sketch-n-etch suggested is a prudent insurance policy. Later designs such as the LF412/442 seem to have more of a margin for error built into their compensation circuits.
Dr. Sketch-n-Etch
gwaidan wrote:
The TL07x/08x series of chips are only just internally compensated for unity gain-throwing an extra compensation cap into the loop like Dr Sketch-n-etch suggested is a prudent insurance policy.

Yes, but don't do it indiscriminately. Sometimes, these feedback capacitors create more harm than good. Putting them in the gain stages of a cascaded filter, for example, actually renders the circuit more unstable. Also, using them when it isn't necessary can create unwanted lags in the system, filtering out high frequencies and decreasing slew rates.

What I tend to do in my own layouts is to make places for stability caps on all the opamps, and then only use them when I find it necessary to do so. It took me three years to get to that point. This is how it goes with electronic design -- it's the accumulation of little tricks and tips, based on hard-won experience over time.
skrasms
Does the short-circuit rating also apply when an op-amp is not powered? I keep the safety resistor (plus a diode clamp) outside the feedback loop for the event that someone connects another output to my outputs while my op-amps aren't powered.
Dr. Sketch-n-Etch
skrasms wrote:
Does the short-circuit rating also apply when an op-amp is not powered? I keep the safety resistor (plus a diode clamp) outside the feedback loop for the event that someone connects another output to my outputs while my op-amps aren't powered.

Honestly, I have no idea. (Hmm....? hmmm..... )
Dave Kendall
This thread is damn useful.....thanks to all posters so far.

Some questions -
1/ Any reason why "innie" *and* "outie" resistors aren't used together (apart from cost/part count)? would they give the benefits of both types, or screw things up?

2/ @Dr S - Is the 4K7 in the feedback loop of the voltage follower there to compensate in some way for the 220r "innie", or another reason? Most voltage followers I've seen have a simple short connection bewteen OUT and - IN.

3/ Dr S again, how critical is the value of the capacitor in the feedback loop?

4/ Many of the 100K 1% I got check out at 99.7 or 99.8 on the DMM. Would using those in the feedback loop along with a 220R "innie" reduce errors for critical CV apps, or doesn't it work like that?

Sorry about all the questions, but buffered outputs are in almost every circuit, so it's well worth knowing lots about the options, so as to decide what approach to take.
Dead cool thread (oh, I said that already...)wink

cheers,
Dave
neil.johnson
skrasms wrote:
Does the short-circuit rating also apply when an op-amp is not powered? I keep the safety resistor (plus a diode clamp) outside the feedback loop for the event that someone connects another output to my outputs while my op-amps aren't powered.


Resistor plus diodes to the rails (one each to rach rail) is a good protection scheme. You can use it in both resistor configurations. Bonus points if you use Schottky diodes.

Neil
neil.johnson
Dr. Sketch-n-Etch wrote:
However, the main difference is in how the opamps respond to downstream capacitive loading. An "outie" resistor provides very good compensation for capacitive loading (and a long patch cable is often enough downstream capacitance to create oscillation problems). And "innie" resistor does nothing for capacitive loading, and in fact can even make matters worse. This is why I apply a special capacitive loading compensation network on all my opamps when I'm serious about not creating CV errors. Basically, my preferred circuits are as shown in the following diagram (note: the 220R "outie" resistors can be 1k)


An "innie" resistor without the compensation capacitor is almost guaranteed to have problems -- you've implemented a half-solution. The correct solution is the resistor plus the compensation capacitor. There's nothing "special" about it -- it's a technique that's as old as the hills and compensates for the pole formed by the load capacitance and the protection resistor.

Neil
neil.johnson
Dr. Sketch-n-Etch wrote:
Concerning CV errors and tuning, yes, most CV circuits in VCOs are based around FET input opamps. However, they are generally organized as inverters, so the input resistor becomes the defacto input impedance. If this resistor is 100k, and the CV source has a 1k output resistor, then there will be roughly a 1% error in the CV, and therefore in the tuning. Using an "innie" resistor eliminates this error.


If feeding the "classic" 100k pot + 100k wiper resistor then the actual input resistance varies between 50k and 100k depending on wiper position. And, as luck would have it, the input resistance goes down, making the error worse (2% for a 1k output resistor), as the wiper is moved towards the input d'oh!

Using a reversed scheme that I wrote up last year minimises this error, but the only way to completely avoid it is to buffer the input before the attenuator pot.

This is another reason why the "innie" circuit is better for CV-type output stages.

Neil
daverj
Dave Kendall wrote:
1/ Any reason why "innie" *and* "outie" resistors aren't used together (apart from cost/part count)? would they give the benefits of both types, or screw things up?


Well, actually it is a third option and does have some advantages, but at the same time you loose other advantages. It's a compromise.

An "outie" that is very small (50 - 75 ohm) will isolate the op amps output from the capacitive load of a cable and prevent the oscillations mentioned earlier. Used in conjunction with an "innie" does give some of the advantages of using the innie, but you lose a bit of that advantage by having an "outie" added to it. Since the "outie" is quite small though the loss is also.

With this method you will still have the output level change with load, but since the output series resistor is so small you lose maybe 0.05% instead of 1% of the signal in a typical load. But you gain the advantage of eliminating capacitive load oscillations without having to restrict the bandwidth of the op amp. And you retain the advantage of having a large enough resistor on the op amp output (the "innie") to limit current and prevent over heating.

The resistor is also too small to do the classic "short two outputs to mix them" trick. But that's generally not a good idea anyways. Mixers are pretty cheap and always a better method.
Navs
Thanks for the illuminating discussion and sharing your knowledge, guys we're not worthy
neil.johnson
Dr. Sketch-n-Etch wrote:
(Now, having said that, I must confess that I really don't understand why this is necessary with TL07X, since it is already internally compensated. I'm not smart enough to understand all the subtleties around this issue, but I do at least know what to do about it when things aren't working properly.)

Op-amps are usually compensated for operation at a particular gain, which is not always x1 (unity). For example, the NE5534 is internally compensated for gains greater than 3. If you want to use it for a unity gain buffer you need to add an external compensation capacitor or some other lead-lag compensation network to ensure stability.

The TL0xx series have internal frequency compensation to make them unity-gain stable, meaning you can hook them up with unity gain without needing to add external compensating capacitors. This is also one of the main reasons for the popularity of the 741 - that damn-great 30pF capacitor saved so much money at the time that it became so easy and cheap to use!!

The downside to this is that it slugs the performance at higher gains since you can't remove the internal capacitor.

Neil
Chrutil
Hi all,
Noob'ish question here. I'm slowly learning, but my 'wants' are still far ahead of my skills. I'm outputting CV from a DAC (ad5668) (this is for a dotcom system) and I understand I'm supposed to protect the circuit, and the trails lead to the CV follower here. I'm settling with 0-5V for now so I shouldn't have any gain, so I was thinking I could use the CV follower below as is.

My question is about the 100pF capacitor - what kind of capacitor is that?

[Edit: ok, by 'should not have any gain' I meant gain should be 1x..]

Many thanks!
C

Dr. Sketch-n-Etch wrote:


Nantonos
A ceramic, and you don't need to pay for super high tolerance either.
Chrutil
ah great! Merci!
MUFF WIGGLER Forum Index -> Music Tech DIY  
Page 1 of 1
Powered by phpBB © phpBB Group