Open Loop Gain Measurement? Fotios Need Help!

Status
Not open for further replies.
I request from any of you has an advice for a PRACTICAL MEANS to measure the open loop gain of an amplification arrangement to give me it.
I have the book (not only the pdf version) of D. Self "Audio Power Amplifiers Design Handbook 2nd edition, for that i paid in the year 2000, 17.400 Drachmas=51Euros of today to by it. I have read it and the method proposed by D. Self it is known to me.
So, i ask for a different method.
Thanks in advance for your polite help in me.

Fotios
 
No one response yet? Why? It is a so difficult issue?
To measure the open loop gain i thing, we must remove the main feedback node? maybe also the miller cap which is a nested feedback loop into the main feedback loop? And i don't know how many other nested and well hidden feedback loops into the main feedback of the amplifier?
By removing only the main feedback node, be it so if we close in a shielded box the amplifier as to be insensitive in external interference, only the touch of the scope probe in the output rail can cause great instabilitty and oscillations.
With which means the great companies make the measurement of the open loop gain?
Have any of the experts in this forum the polite kindness to explain me the instrumentation method used in their laboratories?

Fotios
 
I've made open loop measurements before. I believe it is usually done with nested feedback in place, but global feedback removed, so as to find the best compensation network for global feedback. Even after doing it though, I still found the best solution was empirically determining the global feedback network. Doing it mathematically will definitely give you a good place to start, though. Note that I was doing this for a tube amplifier, a solid state amp may be a little more behaved. I used MATLAB with the simulink toolbox for this, but it can be done otherwise with any linear algebra system (MAPLE I believe is free). To actually measure the open loop gain/phase response of the amp, I just used an oscilloscope, but a vector network analyzer or gain/phase meter would definitely be the best way to go, if you can find a good one for the audio band. A computer sound card with some specialty software may be able to do this function with some attenuation and protection circuitry on the input.
 
I have the new and powerfull software "Multi Instrument Pro-3" of Virtins Technology (compared only with SpectraPLUS and above) installed in my PC (Intel C2D E6850 3GHz, Intel M.B. DQ35JO - 1333 MHz FSB, 8Gb Sync. DDR2 800MHz, Win x64 O.S.) and an external sound card E-MU 0404 (192Ks/s locked internally - 24bit); as you can see there is not any most powerfull system from this (for domestic use at least) at this moment.
Tell me please, the circuit arrangement which is appropriate to make the measurement of the open loop gain.

Thanks a lot
Fotios
 
I have also to express a reasonable query:

What for it is necessary IN PRACTICE the measurement of the open loop gain of an amplifier?

The measurement of the closed loop gain in a definite frequency bandwidth with definite also limits of amplitude variations (i.e. +/-3dB or +/-1dB around an average value) it is not enough?

The measurement of O.L.G. it is most important for academic discussions?

Or it is of big importance for the final behaviour of a given amplifying circuit?

Fotios
 
To measure OLG in practice, you need a sinusoidal swept source of often low voltage to keep the amplifier under test away from clipping. It is important for any amplifier but is sometimes or often overlooked because it can sometimes be an unnecessary step in design. If there are feedback problems such as instability or oscillation, they can often be traced back to open loop response. Knowing the open loop response and the feedback factor (complex) can pinpoint troublesome frequencies where feedback becomes positive, and knowing that one can modify the circuit to (hopefully) correct it. I recommend reading up on phase margin and stability.
 
Firstly thanks a lot for your interest.
It is easy to obtain a fine sinusoidal sweep from 20 to 20000Hz by my very good Hameg function generator of 10MHz and to inject the signal in the input. What must to do further? To disconect the feedback node of amplifier for making the measurement?

Fotios
 
Hi Fotios

It is difficult to measure the open loop gain of a high gain amplifier!

The current view is that the OLG should be measured without disturbing the feedback path.

To measure OLG using feedback components in place, ground the input. If the feedback resistor is Rf and the voltage tap resistor Rg, it is often the case that Rg is connected through a capacitor to keep DC gain low while increasing AC gain. Disconnect this decoupling point (capacitor or resistor) and apply a signal generator to this point instead. The open loop gain is obtained from the output voltage divided by the voltage measured at the feedback tap point.

You will need a sensitive low noise high impedance meter to measure the feedback tap point voltage, which can be around 1 mV for 10V output or so. Normally the gain would be measured with the normal load in place.

Maybe an oscilloscope would work if it did not cause oscillation. Some people have built a differential transistor pair to clip across the feedback point using a jumper wire to make this measurement, but of course this may change the base resistance. But a couple of FETS may work, and give a measureable gain (2 to 5x) which may be enough, even if they have a high distortion (the signal level is low so that even bipolars will be fairly linear- and can be degenerated to make sure).

cheers
John
 
I like to disconnect the nfb, using an external servo to steer the DC operating point. Then I feed the input (with the input LPF, if fitted, bypassed) from a signal generator, viewing both the input sinewave and the output sinewave simultaneously on a dual trace oscilloscope to measure the phase shift and output amplitude as the frequency is increased beyond the audio spectrum (paying attention to the phase shift when the open loop gain drops to the desired closed loop gain)
Measuring the open loop gain at very low frequencies with this method is difficult, as the OLG is usually very high here and the DC servo time constant must be very long not the skew the results, but this doesn’t matter much because, as far as stability criterion is concerned, with a typical solid state amplifier one is only interested in the other end of the spectrum anyway.
This method isn’t extraordinarily precise, but it is quite good enough for empirically selecting the compensation capacitors, which only come in ‘preferred’ values anyway.
It also avoids having to measure small signals from parts of the amplifier that are sensitive to loading.


Cheers,
Glen
 
Sometimes it is very difficult to keep DC stability without the feedback loop in place, a method I have used is to use a large capacitor connected in parallel with the input where the loop is connected, this keeps the loop active for DC but disconnect it for high frequencies, thereby making it possible to measure the open loop response.

Next problem is to measure the loop response high enough, a PC sound card is far from enough as the loop response need to be measured up in the MHz range for many amplifiers. I use a signal generator capable of 20MHz and measure output voltage and phase with an ordinary 200MHz oscilloscope. It is a very slow and tedious method but it is exact enough and possible to use for any DIYer. When measuring these relatively high frequencies it is important to make sure that source and load impedances are as they would be when the amplifier is used otherwise the result can differ.

Regards Hans
 
First of all, many thanks for the time that you spend all of you to reply in a so difficult subject such as the OLG.
You can see that, from yestarday untill now i have only a few replies. Instead for subjects such as output relays and buzzing transformers there is a pactolus of replies in few hours!
From my experience all the methods you propose seems like effective and not dangerous probably to burn any active or passive element of the amplifier because they ensure a stability during the measurement process.
Usually, in my amp. projects i preffer an input sensitivity of 0dBV, or 1Vrms input for full output swing. Also i am busy only with supply levels from +/-60V to +/-86V. Thus as you can understand i am working with a Closed Loop Gain from 38,5dB up to 41,7dB accordingly. Also, because the topologies that i use are of those which guarantee the stability in spite of they offers a supposed superb sound quality, never i have seen phase shift or oscillation of output (yet with the connection of a reactive load the ringing it is small due to 8ìH inductor from 2mm wire which i use in the output) at least up to 100KHz. Above this frequency limit, all the signals are blocked due to the input filtering, the whole feedback loop and the nested feedback loops incorporated.
In reality, i have not any problem in my projects but i am very curious to see which will be the behaviour of one of my amplifiers without the support of all the previous reffered feedback aids.
I like the method proposed by John, because i have an appropriate function generator of 10MHz (Hameg HM8130) which also incorporates a rich menu of settings including also its own output signal voltmeter (i checked this with my DSO, the value indicated it is true) thus there is no need to connect any other instrument in the disconnected feedback node which can cause instability due to its interference with the output of generator.
After i will try the method proposed by Hans which seems likewise good as the John method.
Finally, i will make a comparison between the results of the two methods and i will refer in a post the conclusions.

Thanks a lot for the advices, to all

Fotios
 
Practically speaking about the OLG

I ask myself: Why it is needfull the OLG measurement in practice?
The common logical thought says to me that, the term OLG by alone it does not says nothing. On the contrary, if the term OLG reffered in the stability of an amplification arrangement, it says a very usefull thing. The NFB it is a usefull TRICK to ensure the stability of a circuit. It is the easy way to cover as a blanket the bad behaviour of any element or subcircuit included in the whole amplification circuit. Thus, we have the duty to eliminate each wrong thing in the circuit before we close the feedback loop. The only way to do this, it is the test without the NFB closed. The open loop will reveals any kind of bad behaviour of any element incorporated in the circuit. Thus, we must make any possible effort to eliminate these bad points which cause instability as far as possible, while in the same time we must paid any attention to keep the gain of circuit as big as possible.
Today, with so much ambient interferences caused from any wireless device for example, it is practical impossible to succeed a stability without the NFB trick. Yes, i consider the Feedback as a trick and nothing else.
If we had the possibility to close an amplifier, without feedback, in a lead box for good electromagnetic shielding and to supply it with batteries, maybe the things was by far better and the needed feedback was very little.
Thus i think that the OLG term it is determined better as: The maximum possible gain of an amplifier without feedback with the maximum as possible stability.

Fotios
 
Hi P-A

Thanks again for your advice. From tomorrow and for 2 to 3 days i will try all the suggested methods to find the better and the most easy to setup. As you understand it is not an easy work. I will post my conclusions and the documentation with plots taken from a new circuit which of the practical implementation completted before one hour.

Fotios
 
Hi Fotios

Measuring the OLG under closed loop conditions as I suggested can be tricky if the OLG is very high.

Glen's method of disconnecting the feedback resistor is the "old fashioned" way of making the measurement. One way of achieving this is to split the feedback resistor into two (temporarily) and connect the split point to ground with a very large capacitor. This has to be big enough to give a reasonable measure of the OLG, but as Glen says at very low frequencies it is not so critical.

You can also attenuate the input signal with a 10k-10ohm resistor so that the input voltage is 1/1000 of the normal value. This helps with measurement.

This approach is perhaps the simplest. THe capacitor acts as a low frequency servo - but making this too big may give some LF oscillation.

cheers
John
 
OLG has a measureble and predictable effect on the operating characteristics of the amplifier once the loop is closed. output impedance and distortion are both inversely proportional to Aol/Acl (where Aol is open loop gain and Acl is closed loop gain). for instance, output impedance is Zout=Ro/(Aol/Acl), where Ro is the "raw" output resistance.

the best method to measure Aol would be with a swept (or stepped frequency) sine source and a spectrum analyzer. this would give a plot of Aol vs frequency. with some digital signal generators, it's possible to take a "sweep gate" pulse from the spectrum analyzer and use it to step the signal generator to increment it's frequency. with the spectrum analyzer in peak hold mode, this will produce a gain vs frequency plot automatically. measuring the other parameters such as distortion and output resistance will make closed loop performance predictable.
 
Status
Not open for further replies.