I intend to try and measure the open loop gain the a design using various input stage options over the next couple of days and them relate this to the proposed 26 db closed loop gain.
Excess open loop gain and feedback seem to be the evil curse of these designs.
My guess is that the lower power versions of the diy X Aleph work better with more linear devices like the Jfets and small fets.
The traditional larger mosfet devices seem to work better at higher voltages and currents
Excess open loop gain and feedback seem to be the evil curse of these designs.
My guess is that the lower power versions of the diy X Aleph work better with more linear devices like the Jfets and small fets.
The traditional larger mosfet devices seem to work better at higher voltages and currents
Hi Terry,
Everyone seems more interested in the F5 and talking up design ideas in the X Aleph builders thread.
By coincidence there is some discussion around the effect of the resister divider at the input that seems to be causing some confusion aroud open loop gain.
I started doing some measurements last night and will post a table of results when time permits
As a practical matter doing some measurements is a good way of appreciating the concept of feedback, closed and open loop gain as discussed here.
http://www.diyaudio.com/forums/showthread.php?postid=1475879#post1475879
iMac
Everyone seems more interested in the F5 and talking up design ideas in the X Aleph builders thread.
By coincidence there is some discussion around the effect of the resister divider at the input that seems to be causing some confusion aroud open loop gain.
I started doing some measurements last night and will post a table of results when time permits
As a practical matter doing some measurements is a good way of appreciating the concept of feedback, closed and open loop gain as discussed here.
http://www.diyaudio.com/forums/showthread.php?postid=1475879#post1475879
iMac
Hi Ian,
Really excellent work you're doing on the different input topologies. Very valuable indeed. I look forward to your results.
I also agree completely with your comments regarding the low power Aleph-X amps being built by most people. To be fair though, most people are music lovers and amp builders and not experimenters. They prefer to follow directions and trust others to do the due diligence. Different strokes for different folks. That's what makes a forum such a great place.
Once upon a time I considered using cascoded ZVP3310's as a front end for the AX but decided to go straight to JFET's. And now through your wonderful efforts I get to live through that other alternative vicariously. Marvelous! Great stuff!
Graeme
Really excellent work you're doing on the different input topologies. Very valuable indeed. I look forward to your results.
I also agree completely with your comments regarding the low power Aleph-X amps being built by most people. To be fair though, most people are music lovers and amp builders and not experimenters. They prefer to follow directions and trust others to do the due diligence. Different strokes for different folks. That's what makes a forum such a great place.
Once upon a time I considered using cascoded ZVP3310's as a front end for the AX but decided to go straight to JFET's. And now through your wonderful efforts I get to live through that other alternative vicariously. Marvelous! Great stuff!
Graeme
I bought a whole bunch of tiny crocodile leads today so I can swap out the resesters relatively easily.
Its actually quite interesting to run a series of measurements and then calculate the various gain and feeback values in db.
If there is enough interest I will set up a spreadsheet.
If time permits I will run the 1st series tonight.
iMac
Its actually quite interesting to run a series of measurements and then calculate the various gain and feeback values in db.
If there is enough interest I will set up a spreadsheet.
If time permits I will run the 1st series tonight.
iMac
I did an inital series tonight to get a feel of how to run these measurements.
operating conditions
1k sine wave
Source oscillator unbalanced into Passlabs X2.5 = balanced output
Passlasb X2.5 output voltage level 49mv differential
X Aleph setup
Input devices IRF9610 diff pair
Series 1
R19/29 22K
R16/30 221K
Differential Voltage input voltage at input of R18/28 49mv
Differential Voltage at node of 19/29 and R16/29 = 12mv =12.2 db loss
Differential Voltage at amplifier output
Closed loop = 947mv =25.72 db from input voltage
=37.94 db from node of R19/29 and R16/29
Open loop = 13.48 volts =48.78 db from input voltage
Feedback taken from open less closed loop gain = 23.06 db
Series 2
R19/29 44K
R16/30 221K
Differential Voltage input voltage at input of R18/28 49mv
Differential Voltage at node of 19/29 and R16/29 = 125mv =8.13db gain
Differential Voltage at amplifier output
Closed loop = 959mv =26.00 db from input voltage
=18.8db from node of R19/29 and R16/29
Open loop = 16.28 volts =50.42 db from input voltage
Feedback taken from open less closed loop gain = 24.42db
I propose to check these result tomorrow night and look at 100K and 220K for R19/29 an look at effect of R16/20 for 100K.
I have not conclusions yet. No dout a wider spread of values will so something. We will the look at the other devices ie 2SJ74 and ZVP2210A over the next few days.
iMac
operating conditions
1k sine wave
Source oscillator unbalanced into Passlabs X2.5 = balanced output
Passlasb X2.5 output voltage level 49mv differential
X Aleph setup
Input devices IRF9610 diff pair
Series 1
R19/29 22K
R16/30 221K
Differential Voltage input voltage at input of R18/28 49mv
Differential Voltage at node of 19/29 and R16/29 = 12mv =12.2 db loss
Differential Voltage at amplifier output
Closed loop = 947mv =25.72 db from input voltage
=37.94 db from node of R19/29 and R16/29
Open loop = 13.48 volts =48.78 db from input voltage
Feedback taken from open less closed loop gain = 23.06 db
Series 2
R19/29 44K
R16/30 221K
Differential Voltage input voltage at input of R18/28 49mv
Differential Voltage at node of 19/29 and R16/29 = 125mv =8.13db gain
Differential Voltage at amplifier output
Closed loop = 959mv =26.00 db from input voltage
=18.8db from node of R19/29 and R16/29
Open loop = 16.28 volts =50.42 db from input voltage
Feedback taken from open less closed loop gain = 24.42db
I propose to check these result tomorrow night and look at 100K and 220K for R19/29 an look at effect of R16/20 for 100K.
I have not conclusions yet. No dout a wider spread of values will so something. We will the look at the other devices ie 2SJ74 and ZVP2210A over the next few days.
iMac
I did another series tonight and have included a measurement of Differential Voltage at node of 19/29 and R16/29 under open loop gain conditions
1k sine wave
Source oscillator unbalanced into Passlabs X2.5 = balanced output
Passlasb X2.5 output voltage level 49mv differential
X Aleph setup
Input devices IRF9610 diff pair
Series 1
R19/29 22K
R16/30 221K
Differential Voltage input voltage at input of R18/28 49mv
Differential Voltage closed loop gain at node of 19/29 and R16/29 = 12mv =12.2 db loss
Differential Voltage open loop gain at node of 19/29 and R16/29 = 33mv =3.4 db loss
Differential Voltage at amplifier output
Closed loop = 947mv =25.72 db from input voltage
From node of R19/29 and R16/29 = 38db
Open loop = 13.34 volts =48.69 db from input voltage
From node of R19/29 and R16/29 = 52db
Feedback taken from input less closed loop gain = 22.97 db
Feedback taken from node of 19/29 and R16/29 loop less closed loop gain = 14.2 db
Series 2
R19/29 44K
R16/30 221K
Differential Voltage input voltage at input of R18/28 49mv
Differential Voltage closed loop gain at node of 19/29 and R16/29 = 12mv =12.2 db loss
Differential Voltage open loop gain at node of 19/29 and R16/29 = 39mv =1.979 db loss
Closed loop = 956mv =25.80 db from input voltage
From node of R19/29 and R16/29 = 38.02 db
Open loop = 16.28 volts =50.42 db from input voltage
From node of R19/29 and R16/29 = 52.4db
Feedback taken from input less closed loop gain = 24.62 db
Feedback taken from node of 19/29 and R16/29 loop less closed loop gain = 14.38db
Series 3
R19/29 100K
R16/30 221K
Differential Voltage input voltage at input of R18/28 49mv
Differential Voltage closed loop gain at node of 19/29 and R16/29 = 12mv =12.2 db loss
Differential Voltage open loop gain at node of 19/29 and R16/29 = 43mv =1.130 db loss
Closed loop = 965mv =25.888 db from input voltage
From node of R19/29 and R16/29 = 38.10 db
Open loop = 18.14 volts =51.36 db from input voltage
From node of R19/29 and R16/29 = 52.48db
Feedback taken from input less closed loop gain = 25.48 db
Feedback taken from node of 19/29 and R16/29 loop less closed loop gain = 14.38db
Series 4
R19/29 221K
R16/30 221K
Differential Voltage input voltage at input of R18/28 49mv
Differential Voltage closed loop gain at node of 19/29 and R16/29 = 12mv =12.2 db loss
Differential Voltage open loop gain at node of 19/29 and R16/29 = 46mv =0.5469 db loss
Closed loop = 967mv =25.90 db from input voltage
From node of R19/29 and R16/29 = 38.12 db
Open loop = 19.25 volts =51.64 db from input voltage
From node of R19/29 and R16/29 = 52.42db
Feedback taken from input less closed loop gain = 25.74 db
Feedback taken from node of 19/29 and R16/29 loop less closed loop gain = 14.30db
1k sine wave
Source oscillator unbalanced into Passlabs X2.5 = balanced output
Passlasb X2.5 output voltage level 49mv differential
X Aleph setup
Input devices IRF9610 diff pair
Series 1
R19/29 22K
R16/30 221K
Differential Voltage input voltage at input of R18/28 49mv
Differential Voltage closed loop gain at node of 19/29 and R16/29 = 12mv =12.2 db loss
Differential Voltage open loop gain at node of 19/29 and R16/29 = 33mv =3.4 db loss
Differential Voltage at amplifier output
Closed loop = 947mv =25.72 db from input voltage
From node of R19/29 and R16/29 = 38db
Open loop = 13.34 volts =48.69 db from input voltage
From node of R19/29 and R16/29 = 52db
Feedback taken from input less closed loop gain = 22.97 db
Feedback taken from node of 19/29 and R16/29 loop less closed loop gain = 14.2 db
Series 2
R19/29 44K
R16/30 221K
Differential Voltage input voltage at input of R18/28 49mv
Differential Voltage closed loop gain at node of 19/29 and R16/29 = 12mv =12.2 db loss
Differential Voltage open loop gain at node of 19/29 and R16/29 = 39mv =1.979 db loss
Closed loop = 956mv =25.80 db from input voltage
From node of R19/29 and R16/29 = 38.02 db
Open loop = 16.28 volts =50.42 db from input voltage
From node of R19/29 and R16/29 = 52.4db
Feedback taken from input less closed loop gain = 24.62 db
Feedback taken from node of 19/29 and R16/29 loop less closed loop gain = 14.38db
Series 3
R19/29 100K
R16/30 221K
Differential Voltage input voltage at input of R18/28 49mv
Differential Voltage closed loop gain at node of 19/29 and R16/29 = 12mv =12.2 db loss
Differential Voltage open loop gain at node of 19/29 and R16/29 = 43mv =1.130 db loss
Closed loop = 965mv =25.888 db from input voltage
From node of R19/29 and R16/29 = 38.10 db
Open loop = 18.14 volts =51.36 db from input voltage
From node of R19/29 and R16/29 = 52.48db
Feedback taken from input less closed loop gain = 25.48 db
Feedback taken from node of 19/29 and R16/29 loop less closed loop gain = 14.38db
Series 4
R19/29 221K
R16/30 221K
Differential Voltage input voltage at input of R18/28 49mv
Differential Voltage closed loop gain at node of 19/29 and R16/29 = 12mv =12.2 db loss
Differential Voltage open loop gain at node of 19/29 and R16/29 = 46mv =0.5469 db loss
Closed loop = 967mv =25.90 db from input voltage
From node of R19/29 and R16/29 = 38.12 db
Open loop = 19.25 volts =51.64 db from input voltage
From node of R19/29 and R16/29 = 52.42db
Feedback taken from input less closed loop gain = 25.74 db
Feedback taken from node of 19/29 and R16/29 loop less closed loop gain = 14.30db
Hi Ian,
Thank you for publishing this work. It really helps us all. As for the mysterious 10dB, my recollection is that it wasn't a sacred or hard and fast thing. And I don't remember it being used in reference to the XA series.
The numbers you are uncovering are in the ball park of what I would have expected to see for the XA/AX circuit. I'm really interested in seeing what you measure for the dual JFET input version.
Graeme
Thank you for publishing this work. It really helps us all. As for the mysterious 10dB, my recollection is that it wasn't a sacred or hard and fast thing. And I don't remember it being used in reference to the XA series.
The numbers you are uncovering are in the ball park of what I would have expected to see for the XA/AX circuit. I'm really interested in seeing what you measure for the dual JFET input version.
Graeme
Graeme,
Yes it will be.
I would like to see more input so here is some food for thought...
A lot of people perhaps think these older X Aleph are a bit lame compared to the more contemporary First Watt designs .
A lot of them were built, scaled and cloned on the basis they will just go which is what we want to see but as a rule the focus has been on offset voltages.
Pity I don't have a distortion analyser.
I must look into that because I am coming around the the realisation that all these simple 2 stage designs are set up for best measured distortion and sound quality.
A bit off the best vaues and you may have possibly 10X the best distortion. Of course the caveats of voltage and bias always apply.
From what little I have done so far adjusting the feedback divider makes quite a difference
I also suspect the actual input pair bias is the kicker (apart from the device selection) if one is looking to create a given closed and open loop gain scenario. I plan to vary that within reason.
A lot of peole worry about driving the gate capacitance of the followers. Just look at the curves in Nelson's article on measuring Fets. Just pour on the voltage and up the current of those big suckers. If you must use 15 volts do so at your own peril.
I am not trying to stir the pot here but without stating the obvious don't assume too much with any of this stuff.
Forgot to do 10K resister to ground and will look at that tonight and then 100K feeback resisters.
I will then provide a matrix comparing the open and closed loop gain if the Jets and the small ZVP2110
Yes it will be.
I would like to see more input so here is some food for thought...
A lot of people perhaps think these older X Aleph are a bit lame compared to the more contemporary First Watt designs .
A lot of them were built, scaled and cloned on the basis they will just go which is what we want to see but as a rule the focus has been on offset voltages.
Pity I don't have a distortion analyser.
I must look into that because I am coming around the the realisation that all these simple 2 stage designs are set up for best measured distortion and sound quality.
A bit off the best vaues and you may have possibly 10X the best distortion. Of course the caveats of voltage and bias always apply.
From what little I have done so far adjusting the feedback divider makes quite a difference
I also suspect the actual input pair bias is the kicker (apart from the device selection) if one is looking to create a given closed and open loop gain scenario. I plan to vary that within reason.
A lot of peole worry about driving the gate capacitance of the followers. Just look at the curves in Nelson's article on measuring Fets. Just pour on the voltage and up the current of those big suckers. If you must use 15 volts do so at your own peril.
I am not trying to stir the pot here but without stating the obvious don't assume too much with any of this stuff.
Forgot to do 10K resister to ground and will look at that tonight and then 100K feeback resisters.
I will then provide a matrix comparing the open and closed loop gain if the Jets and the small ZVP2110
Series 5
R19/29 10K
R16/30 221K
Differential Voltage input voltage at input of R18/28 49mv
Differential Voltage closed loop gain at node of 19/29 and R16/29 = 12mv =12.2 db loss
Differential Voltage open loop gain at node of 19/29 and R16/29 = 24mv =6.19 db loss
Closed loop = 918mv =25.45 db from input voltage
From node of R19/29 and R16/29 = 37.67db
Open loop = 9.51 volts =45.75 db from input voltage
From node of R19/29 and R16/29 = 51.95db
Feedback taken from input less closed loop gain = 20.3 db
Feedback taken from node of 19/29 and R16/29 loop less closed loop gain = 14.28db
Series 6
R19/29 10K
R16/30 100K
Differential Voltage input voltage at input of R18/28 49mv
Differential Voltage closed loop gain at node of 19/29 and R16/29 = 12mv =12.2 db loss
Differential Voltage open loop gain at node of 19/29 and R16/29 = 24mv =6.19 db loss
Closed loop = 439mv =19.03 db from input voltage
From node of R19/29 and R16/29 = 31.26
Open loop = 9.51 volts =45.75 db from input voltage
From node of R19/29 and R16/29 = 51.95db
Feedback taken from input less closed loop gain = 26.72 db
Feedback taken from node of 19/29 and R16/29 loop less closed loop gain = 20.69db
R19/29 10K
R16/30 221K
Differential Voltage input voltage at input of R18/28 49mv
Differential Voltage closed loop gain at node of 19/29 and R16/29 = 12mv =12.2 db loss
Differential Voltage open loop gain at node of 19/29 and R16/29 = 24mv =6.19 db loss
Closed loop = 918mv =25.45 db from input voltage
From node of R19/29 and R16/29 = 37.67db
Open loop = 9.51 volts =45.75 db from input voltage
From node of R19/29 and R16/29 = 51.95db
Feedback taken from input less closed loop gain = 20.3 db
Feedback taken from node of 19/29 and R16/29 loop less closed loop gain = 14.28db
Series 6
R19/29 10K
R16/30 100K
Differential Voltage input voltage at input of R18/28 49mv
Differential Voltage closed loop gain at node of 19/29 and R16/29 = 12mv =12.2 db loss
Differential Voltage open loop gain at node of 19/29 and R16/29 = 24mv =6.19 db loss
Closed loop = 439mv =19.03 db from input voltage
From node of R19/29 and R16/29 = 31.26
Open loop = 9.51 volts =45.75 db from input voltage
From node of R19/29 and R16/29 = 51.95db
Feedback taken from input less closed loop gain = 26.72 db
Feedback taken from node of 19/29 and R16/29 loop less closed loop gain = 20.69db
I have been reading Opamps Explained by Bryan Maher.
It is perhaps useful to point out that with opamps that have low open loop gain G, there will be some error in T = the closed loop gain where H = R1/R1+R2 as expressed here:
Closed loop gain T= G/ I + GH. In opamps where G is so large that the I in the demoninator is insignificant, hence T = G/Gh approx.
As an example a modern opamp with G =30,000 and where T = 25, the voltage divider H= R1/(R1+R2) = 1/25
In our example where G may only equal x 194 and we want H = 10,000 / (10,000+100,000) = 0.0909 =11 There may be differences in calaculated and measured gains because G falls with frequency and because of gain accuracy errors where G is a relatively small number where T= G/l+GH
To give some idea in the above reference there is a table that ilustrates the errors.
Where R1 = 1k and R2 = 99K H= 1/100.
For an open loop gain of G = 10 and where nominal closed loop gain is 100, actual gain T= 9.09
The situation improves with G =100, where nominal closed loop gain is 100 , actual gain T = 50.00
For G=1000 , where nominal closed loop gain is 100 , actual gain T =90.90
True gain accurancy is not obtained until G=10,000 in this example.
The situation is improved somewhat where ;
R1 = 1k and R2 = 2K where h = 1/3
For G = 100 , nominal closed loop gain of 3, actual T= 2.91
For G = 1000, nominal closed loop gain of 3, actual T = 2.99.
Therefore expect some degree of inaccuracy between computed and actual measured values of these low G amplifiers.
It is perhaps useful to point out that with opamps that have low open loop gain G, there will be some error in T = the closed loop gain where H = R1/R1+R2 as expressed here:
Closed loop gain T= G/ I + GH. In opamps where G is so large that the I in the demoninator is insignificant, hence T = G/Gh approx.
As an example a modern opamp with G =30,000 and where T = 25, the voltage divider H= R1/(R1+R2) = 1/25
In our example where G may only equal x 194 and we want H = 10,000 / (10,000+100,000) = 0.0909 =11 There may be differences in calaculated and measured gains because G falls with frequency and because of gain accuracy errors where G is a relatively small number where T= G/l+GH
To give some idea in the above reference there is a table that ilustrates the errors.
Where R1 = 1k and R2 = 99K H= 1/100.
For an open loop gain of G = 10 and where nominal closed loop gain is 100, actual gain T= 9.09
The situation improves with G =100, where nominal closed loop gain is 100 , actual gain T = 50.00
For G=1000 , where nominal closed loop gain is 100 , actual gain T =90.90
True gain accurancy is not obtained until G=10,000 in this example.
The situation is improved somewhat where ;
R1 = 1k and R2 = 2K where h = 1/3
For G = 100 , nominal closed loop gain of 3, actual T= 2.91
For G = 1000, nominal closed loop gain of 3, actual T = 2.99.
Therefore expect some degree of inaccuracy between computed and actual measured values of these low G amplifiers.
Hi Ian,
Great work !
If it can help you, I once did the maths for the gain of an ugs stage, and the same formula should apply here. It includes the effets of the input to ground resistor. and the feedback network as well.
It's here :
http://psykok.homelinux.org/diy/preamp_ugs/pdf/UGS pour les nuls.pdf
Sorry, it's in french, but the maths should be sufficient per se.
Look at page 13 & 14, and eq 32, where Rfb is the feedback resistor, Rin the input resistor, Rg the ground resistor and Ad the differential open loop gain.
On the last pages, you can see how the variarion of a resistor value impacts the gain ("entree" means input)
Hope this helps, and keep up this good work
Cheers
Great work !
If it can help you, I once did the maths for the gain of an ugs stage, and the same formula should apply here. It includes the effets of the input to ground resistor. and the feedback network as well.
It's here :
http://psykok.homelinux.org/diy/preamp_ugs/pdf/UGS pour les nuls.pdf
Sorry, it's in french, but the maths should be sufficient per se.
Look at page 13 & 14, and eq 32, where Rfb is the feedback resistor, Rin the input resistor, Rg the ground resistor and Ad the differential open loop gain.
On the last pages, you can see how the variarion of a resistor value impacts the gain ("entree" means input)
Hope this helps, and keep up this good work
Cheers
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.