New Posts New Posts RSS Feed - Inside op-amps
  FAQ FAQ  Forum Search   Events   Register Register  Login Login

Welcome to the Graham Slee Audio Products Owners Club

 

Open to all owners plus those contemplating the purchase of a Graham Slee audio product wishing to use our loaner program: join here
Subscribe to our newsletter here (Rules on posting can be found here)


Inside op-amps

 Post Reply Post Reply Page  <1 910111213 14>
Author
Graham Slee View Drop Down
Admin Group
Admin Group
Avatar
Telling it as it is

Joined: 11 Jan 2008
Location: South Yorkshire
Status: Offline
Points: 13452
Post Options Post Options   Thanks (2) Thanks(2)   Quote Graham Slee Quote  Post ReplyReply Direct Link To This Post Posted: 03 Aug 2022 at 10:58am
The difficulty in using decompensated op-amps in an RIAA stage, is that unity gain is reached at some point. If not unity, then often a lower gain than 3 or 5 WRT the NE5534 and OP37.

These are not the only op-amps featuring decompensation, so watch out for so called "recommendations", and get to know op-amps (they're not like valves).

Experience shows that phase which is made to "kick back up" to preserve margins, are the op-amps that sound poor. However, that doesn't seem to put off the "rollers". Ignorance is bliss...

Headphone amps don't need much voltage gain, often less than 3 or 5 is quite sufficient, so unless you're able to work out the poles and zeros that lead to a stable noise gain, steer clear.

Basically, three stage op-amps are going to give you grief. There might be the odd exception, such as a rail to rail device, but is it suited to audio? Op-amps are not exclusively made for audio, but if all you ever do is frequent audio forums, perhaps you'd never know?

As an example, one op-amp application I worked on was a HGV rolling road, used to test brakes on tractor units as well as trailers. These op-amps were at times used for audio by some, and they have to go right out to high frequencies to be able to have high voltage gain at DC. These too had to be compensated, otherwise the readings on the meters would be in error. They amplify a tiny voltage from a strain gauge bridge, which shows braking effectiveness.

Other uses might be in temperature measurement, and also in medical electronics. Op-amps are often designed with a specific main use, but the manufacturer will also want to sell it into other industries.

A decompensated op-amp could be compensated for RIAA use by adding resistance in series with the 2122Hz pole capacitor, but that seriously affects the 20kHz gain, so must be compensated for after Riso, using an RC filter to continue the roll-off. The problem here is the overload margin has to be increased to cater for it.

This is why the experienced designer has to way up all things, and choose the most appropriate op-amp. If the designer has to redesign a product to use a different op-amp, then component values change. So, if somebody wanted to "roll in" an AD711 to an AD817 designed Era Gold V, or an AD817 into an AD711 designed Era Gold V, then there will be a noticeable difference in the bass "quantity" as well as quality. If then that somebody preferred one above the other, then that somebody ought to seriously consider tone controls!!!
Not simple enough for Google-Bot to understand...
Back to Top
Graham Slee View Drop Down
Admin Group
Admin Group
Avatar
Telling it as it is

Joined: 11 Jan 2008
Location: South Yorkshire
Status: Offline
Points: 13452
Post Options Post Options   Thanks (1) Thanks(1)   Quote Graham Slee Quote  Post ReplyReply Direct Link To This Post Posted: 04 Aug 2022 at 6:45am
Passive RIAA Design

On a 30V rail an op-amp may be specified to deliver 24V p-p, that's 12V peak, which is 8.4V r.m.s. Much depends on the load it's given.

A 5mV magnetic cartridge playing an RIAA LP record will output 50mV at 20kHz, and a 10mV magnetic cartridge at 20kHz outputs 100mV.

We know that between 500Hz and 2500Hz the groove velocity can increase up to 5 times before groove breakover, but at 20kHz we know nothing. Do we speculate or do we accept the above outputs as being maximum?

We could conjecture using what we know about high frequency groove "wiggle" that it gets tighter and tighter with increasing frequency, especially on the inner end of the groove. And that if modulation were to be 5 times, it might simply just breakover constantly.

So what overload value should we give it? Well, there is groove damage because PVC is relatively soft, and readily accepts little cuts from handling, and what is the width of these cuts relative to the HF "wiggle"?

OK, perhaps we stick to 5 times overload throughout the frequency range.

Therefore, the maximum is going to be 100mV * 5 or 500mV at 20kHz, through considered assumption.

The maximum gain of a first stage passive RIAA should therefore be 8.4V / 0.5V = 16.8. And we could EQ the highs after this stage.

For the low frequency range we have another 20dB to EQ, and so again the maximum gain is 16.8.

In effect, with a 5mV cartridge, we start at 0.5mV and apply two lots of gain at 16.8 times each, which is 282, so the output after two passive EQ stages is 141mV.

The last passive filter is unable to drive an "outside world" input such as an amplifier input, so must be buffered, and we may as well apply some gain to get a little more output than 141mV.

Now, with a 10mV cartridge, the output will be 282mV, and if we add the 5 times overload, that's 1.41 volts into this last stage. And the maximum is 8.4V, so 8.4/1.41 = 6.

We multiply the gains and divide by 10 to find the output for different cartridges. The gain is 169.344. There might be insertion losses which we haven't discovered yet, so decimal point accuracy seems silly.

The output for a 10mV cartridge is 1.69 volts - quite impressive! Hope you were following?

After the first stage there's a simple RC filter turning over at 2122Hz. May as well use a 7.5k resistor and a 10nF capacitor. The time constant is

7500 * 0.000,000,01 = 0.000075 = 75us (correct!)

After the second stage there's a RCR filter turning over at 50Hz, and shelving at 500Hz, so that's a 50Hz pole and a 500Hz zero.

Now here is where the passive EQ can get a little bit unstuck. There isn't much charge and discharge current available, so the capacitor needs to have the lowest dielectric absorption we can find. Oh, and its inductance, well it might not have much inductance but we don't have that much current to overcome it.

If it's wound, it's a coil, it has inductance, wool over eyes - do not pull!

All metal film capacitors are wound because there's no other economic way of doing it, but if slit at 0 and 180 degrees, pressed, trimmed and ends welded, you have slit foil stacked film, which is as inductive as the wires joining it (not very).

Now, what about its dielectric absorption (DA)? Yes, insulation can store charge, especially plastics. Wear a nylon shell suit and you'll be discharging all over the place!

We want something inert, non stick, but Teflon is very expensive, and difficult to find. Now air would be a great dielectric, but suffers with pollution. Let's try polypropylene. Provided it's not "wound".

The limit seems to be 10nF, as any larger it loses its characteristics. So it's 10nF again.

So, at 50Hz we need a series resistor of 318 kilohms. Hmm, that's going to have a noise voltage of 1,272 nV/√Hz.

And this is low frequency noise, increasing with the inverse of frequency, and then we have the joys of 1/f noise at low frequency.

Add to that the fact that we have three non-inverting op-amps following each other and they all have common mode distortion.

So, what's any good about a passive RIAA?

At this point it isn't worth designing.
Not simple enough for Google-Bot to understand...
Back to Top
Graham Slee View Drop Down
Admin Group
Admin Group
Avatar
Telling it as it is

Joined: 11 Jan 2008
Location: South Yorkshire
Status: Offline
Points: 13452
Post Options Post Options   Thanks (1) Thanks(1)   Quote Graham Slee Quote  Post ReplyReply Direct Link To This Post Posted: 04 Aug 2022 at 8:07am
Fantastic op-amps with one massive flaw

The LF156 (356) and LT1056 feature all the great things you'd ever need in audio, except equal slew-rate.

Mr Jung puts it down to the p-channel input stage, and he's probably right, but the output stage - complicated by short circuit proofing - is almost the NE5534 output stage (spot the Shaw diode!).

Quasi complimentary outputs like these also have unequal positive and negative slew rates.


As long as they're made to behave (using "VCCS GOSIT" in simulating the case of the 156/1056), as output stages quasi's can sound great, but you'd not be advised to use a quasi where capacitors are used for EQ.

The charge rates differ because the slew rates differ.

The audible effects can be minimised by using low DA capacitors, which is why the NE5532 Gram Amp 2 trounced the cheaper NE5532 Gram Amp 1 (the Gram Amp 1 used polyesters).

The LF411, however, has near equal slew rate, and just about right for EQ such as RIAA, provided it's high enough impedance (not much output current). Again, the downside is there is no boasting about S/N, but oh, the sound!
Not simple enough for Google-Bot to understand...
Back to Top
Graham Slee View Drop Down
Admin Group
Admin Group
Avatar
Telling it as it is

Joined: 11 Jan 2008
Location: South Yorkshire
Status: Offline
Points: 13452
Post Options Post Options   Thanks (0) Thanks(0)   Quote Graham Slee Quote  Post ReplyReply Direct Link To This Post Posted: 05 Aug 2022 at 5:19am
Op-amps can work on a single supply too! Or can they?

Back in 1974 they were considered to be able to do anything! Yes, just use three "blocking capacitors", input, output, and at Rg - perfect performance every time!

A few years ago ADI produced an application note showing the correct way of using op-amps on a single supply. This was in some ways a sales gimmick. OK, it works, sort of. It was to show designers of battery portable items how to utilise op-amps.


Basically, an op-amp is a super high gain power amplifier in miniature, etched and deposited onto a sliver of silicon (a chip or dice - as in diced - sometimes called die).

It's housing is called a leadframe (lead as in leadouts). The chip has connection lands, and the lands are connected to the leadframe pins by tiny gold wires. They are gold because gold is inert where other metals suffer dissimilar metal corrosion. The wires are "spot welded" either end - just think what would happen if they were soldered...!

After that, they are encapsulated, and the frame cut away from the pin ends.

What if you made a single HT power amplifier with only one ground?

Let's take that a stage further: what if you made a stereo single HT power amplifier with just one ground?

The negative terminal is zero volts on a single supply. The inputs and outputs are held around mid voltage for there to be a symmetrical signal swing. Their DC is blocked by input, output and Rg capacitors.

The terminal the gain depends on has also output current flowing in it, but worse than that, it is half wave "rectified" due to how the output capacitor charge is held.

But surely, it is of such tiny current that it should not matter? That's how it was with valve amplifiers and they sounded alright. It was only when low impedance transistor amplifiers came into being that the current was more substantial.

Yes, but valves are much less complicated, and have less gain by comparison.

Once this is realised, the single supply designer might wish to avoid op-amps altogether, and employ discrete design. However, discrete design is a lot more demanding than op-amps. Today, many designers of low level circuits are really op-amp orientated having not done anything discrete. These are well advised to steer clear of single supplies.

However, if you're forced into a corner by circumstances, and have to work on single supply op-amp circuits, the downsides have to be mitigated.

Would you therefore want a three stage op-amp? Would a simpler op-amp mitigate the situation? Perhaps take advantage of EN 61938 to reduce the output load, yet still make products within the matching system?

You would also need to ensure its power supply was optimised for this particular problem.

Yes, I am a single supply designer. Would any clever arse like to have a go?

I will conclude my contributions to this topic now, but I will be happy to answer any questions you feel are important to you.
Not simple enough for Google-Bot to understand...
Back to Top
Ash View Drop Down
Senior Member
Senior Member
Avatar

Joined: 18 Mar 2013
Location: Dorset
Status: Offline
Points: 3983
Post Options Post Options   Thanks (0) Thanks(0)   Quote Ash Quote  Post ReplyReply Direct Link To This Post Posted: 05 Aug 2022 at 7:32am
Taking your Voyager headphone amplifier as an example then, it uses a 9V PP3 battery. Does the circuit use this supply as +4.5V and -4.5V?

Same with the Solo. The 24V is actually used as +12V and -12V?


Edited by Ash - 05 Aug 2022 at 7:34am
A person convinced against their will is of the same opinion still.
Back to Top
Graham Slee View Drop Down
Admin Group
Admin Group
Avatar
Telling it as it is

Joined: 11 Jan 2008
Location: South Yorkshire
Status: Offline
Points: 13452
Post Options Post Options   Thanks (1) Thanks(1)   Quote Graham Slee Quote  Post ReplyReply Direct Link To This Post Posted: 05 Aug 2022 at 9:04am
The reason for the outboard DC power supply was to comply with the legislative directives to ensure the legality of my products.


The single market economy of Europe and since Brexit, the same standards are adopted by UKCA, requires mains powered equipment to meet certain standards, and it is the manufacturers responsibility to ensure safety.

An outboard power supply approach, where that power supply meets with safety standards, is often used, and sometimes the responsibility is passed, third party to the power supply manufacturer.

It is very difficult to circumvent the law.

Not that I want to circumvent it, I actually proved the PSU1 through a certified international standards body: Kema Quality B.V., in the Netherlands. As such, the power supply of our own manufacture was certified as compliant with CE marking and that also meets with UKCA.

However, in the EC and UK, and many other countries, the manufacturer can self certify the safety of its mains powered products by declaring the relevant standard has been applied. I got my certification to please a complaining Canadian distributor.

The items thus powered become Separated Extra Low Voltage (SELV), which means they're safe? They're only safe electrically if they do not compromise the external power supply. There are other safety aspects such as sharp corners etc.

A plug-top transformer might be used to provide a.c. to an a.c. powered SELV item. However, that combination must meet with the emissions part of EMC law.

If the SELV item contains a CIF rectifier (capacitive input filter) then substantial harmonics exist between the SELV item and the plug-top transformer, and the (usually) 5 foot bell wire low voltage cable becomes a transmitter of the harmonics.

That contravenes EMC emissions legislation.

The rectifier switches transformer current to the CIF for approximately one sixth of each half cycle. If half a cycle is classed as 100Hz, the switching lasts for approximately 600Hz per 100Hz to put it crudely.

The plug top transformer's primary sees two inductances, its on-load inductance and its off-load inductance. The eddy current effect sets up a kind of oscillatory action described as harmonic current.

The approx. 5ft wires (or any length for that matter) are a continuation of the transformer's secondary. The harmonic pulses travel along these wires and will radiate as an electric field, which is an emission.

Such a.c. power supplies only meet EMC when supplying an a.c. load. A CIF rectifier filter is not a purely a.c. load.

The use of a DC external supply to power DC apparatus, does not emit these harmonics, because the harmonic path is contained within the power supply unit. If the paths from transformer via rectifier to the CIF are short, the harmonic current cannot develop much if any emissions in its internal wiring.

The use of an E-I laminated transformer within the power supply assists the reduction in harmonics due to its primary magnetising current "swamping" the inductive effect on magnetising current, acting as a brake on transformer harmonics.

As such, the DC power supply is better able to comply with emission levels than an a.c. supply with a CIF load.

------------------------------------------------------------

Reference: EFFECT OF TRANSFORMER LEAKAGE INDUCTANCE
Not simple enough for Google-Bot to understand...
Back to Top
Graham Slee View Drop Down
Admin Group
Admin Group
Avatar
Telling it as it is

Joined: 11 Jan 2008
Location: South Yorkshire
Status: Offline
Points: 13452
Post Options Post Options   Thanks (1) Thanks(1)   Quote Graham Slee Quote  Post ReplyReply Direct Link To This Post Posted: 05 Aug 2022 at 11:08am
Originally posted by Ash Ash wrote:

Taking your Voyager headphone amplifier as an example then, it uses a 9V PP3 battery. Does the circuit use this supply as +4.5V and -4.5V?

Same with the Solo. The 24V is actually used as +12V and -12V?

Ash,

An op-amp, if it were a thinking being, wouldn't know if it had +/- or just +.

single-v-dual rail

It doesn't have a pin marked 0V, just positive and negative - a bit like a torch bulb. An EMF is supplied across it and if it requires 9 volts, then it's either +9V with respect to the negative terminal, or -9V with respect to the positive terminal.

If you want it to output a signal placed such that the signal is centred between zero and +9V, you make the output +4.5V.

You can do that by two equal resistors in series across the 9V, and connect the midpoint to its non-inverting input, and provided they match the feedback resistor, the op-amp will sit at +4.5V.

As an example of "match": if the feedback resistor is 100k ohms, the two resistors are 200k ohms each. Why? because the two resistors of a potential divider are effectively in parallel.

Their centre point is mathematically 4.5V, the op-amp output is 4.5V, the inputs won't be because they require bias current, and the same bias current is used by each input, so each input voltage is the same. With a high bias current input, the input voltage might be several millivolts up or down depending on the direction of current flow to/from the inputs, or with a low bias current, you'd probably swear it was 4.5V.

Or,

You can use a split supply that has a mid rail which you call 0V, and just "drop" one 100k resistor to the non-inverting input.

Perhaps your teacher had other things on his/her/its mind? Was it a mixed class, or perhaps it was politics....

The problems start when you've to divide voltages to obtain gain. With a mid rail you drop say a 10k resistor from inverting input to the mid rail (0V) to get a gain of 11 ( 1+(100k/10k) ), and then you notice the output has offset one way or the other.

So then you have to make the resistance to the other input have 100k in parallel with 10k (or a 9.1k resistor), but then 9.1k is a too heavy load for your input signal - you wanted 100k.

OK then, that's easy. Break the current path between the 10k and 0V with a capacitor. The op-amp then reverts to a 0V (still really 4.5V) output.

But often, the capacitor needs to be a large value which means electrolytic, and these should be polarised right way up.

Yes, but all capacitors are bad, so we'll make a rod for our backs and use another op-amp to make a servo to force first op-amp to do what it really didn't want to (this part not shown in diagram).



Edited by Graham Slee - 05 Aug 2022 at 11:12am
Not simple enough for Google-Bot to understand...
Back to Top
 Post Reply Post Reply Page  <1 910111213 14>
  Share Topic   

Forum Jump Forum Permissions View Drop Down

Forum Software by Web Wiz Forums® version 12.01
Copyright ©2001-2018 Web Wiz Ltd.

This page was generated in 0.078 seconds.