The Asus Xonar U7 was not designed as a measuring instrument. Although not bad DAC and ADC are installed in it, this is a multi-channel gaming card with all the ensuing consequences.
LME49726 with 300mA output and 0.7mA @ 5V quiescent current is good for headphones. LME49726 is not suitable for ADC buffering.
Last edited:
Hello everyone once again 
I Just leave here info, cause this is the best thread about upgrading Rotel.
I have 2 info
1. So, i was playing with the Ra-01 model and finally have decided to modify power amp. I have add there 4 capacitors 1000uF (blue) and 4 resistors 10Ohm(red). The effect is just WOW. Highly recommended. Just audible recommendation, no measurement.
2. If anyone of you, will seek for main power caps, so I have tried many... Rotel Ra01, RA02 are very thin so no many choices. But I've found the best, I think. I recommend Epcos B41231 - you can get 8200 uF. Mundorf caps could be only 6800uF, but they mostly are much less than should be. You will be Lucky if you get 6200 uF.
Could you recommend me $ affordable soundcard to use Right Mark Audio Analyzer? Or how you do this measurements of THD, etc?
Black line in the drawing is the common wire that leads to ground. Those 2 capacitors in reality are next to each other and have common grounding wire.
I Just leave here info, cause this is the best thread about upgrading Rotel.I have 2 info
1. So, i was playing with the Ra-01 model and finally have decided to modify power amp. I have add there 4 capacitors 1000uF (blue) and 4 resistors 10Ohm(red). The effect is just WOW. Highly recommended. Just audible recommendation, no measurement.
2. If anyone of you, will seek for main power caps, so I have tried many... Rotel Ra01, RA02 are very thin so no many choices. But I've found the best, I think. I recommend Epcos B41231 - you can get 8200 uF. Mundorf caps could be only 6800uF, but they mostly are much less than should be. You will be Lucky if you get 6200 uF.
Could you recommend me $ affordable soundcard to use Right Mark Audio Analyzer? Or how you do this measurements of THD, etc?
Black line in the drawing is the common wire that leads to ground. Those 2 capacitors in reality are next to each other and have common grounding wire.
Attachments
It is mandatory to install such a filter (10 Ohm 1000uF = 15 Hz at -3 dB level). With a decline of - 6 dB per octave, the calculated decline starts from 20 Hz. I have a calculated 15 Hz, but 33 ohms 470 microfarads. Developers usually take into account the decline not by an octave, but by a decade, or a stabilized power supply for the differential stage and voltage amplifier is necessary. For a larger capacitance of the capacitor, I'm afraid there is not enough space in Rotels, and an increase in resistance will lower the supply voltage, which is not good.
The second most important is the breeding of the board, but here it is more difficult.
Without modifications motu m4, with modification Asus Xonar U7 MKII.
The cure for this problem seems to me to be ground with a star and with a case at the input connectors, and not like in Rotel in the center of the board. Whether it is possible to implement it in Rotel I did not try. Who tried to write.
Hi Per,
The signal ground (differential stage ground) was connected by a separate conductor at a common point. This greatly reduced mains harmonics and greatly improved the sound quality.
Hi do you suggest to by-pass the line stage entirely ? i understand that the best rated Rotel preamps (and more expensive like the one of the old Michi line) are those using discrete components and not opamps Line preamps i mean
I have removed the preamp in my Rotel RA-972. The input signal from the connector is fed directly to the input of the differential stage of the power amplifier via a switched source select relay. In this case, the input grounds of the RCA, the volume control and the bases of the transistors of the differential stage of the power amplifier are connected by separate conductors at a common point.
hi thank you for the very kind and helpful advice
so you are using it as a power amp with a separate preamp?
i have read some reviews about rotel pre plus power amp models
the opamp based preamps have received so so rating Instead the power amps have been judged all positively
the totl Rotel preamps are all discrete
imho to find a very good line preamp is more difficult than to find a good power amp
i don't know why
I don't have a preamp, just a volume control.
21st century and 2V RMS is the output standard, so 20th century reviews are outdated. Amplifiers Rotel input 1V RMS. Why the extra amplifying stage? After all, the shorter the path, the better. Also, a 50 kOhm potentiometer does nothing except increase the noise. All modern sources are capable of working with 10 kOhm, unless of course you are a fan of tube technology.
At the source output, I have OPA1656 100mA enough.
Last edited:
Thank you very much again Very helpful indeed I like tubes but i am not a fan The problem is that i like to try different power amps (or even active speakers recently) and a line preamp able to drive almost anything is handy But i believe in your approach Sometimes integrateds have too many gain stages What is the reason for all this high open loop gain to reduce it in the end ? Thanks a lot again
Switching works for me in full, that is, the input relay and indication are intact. The power supply of these circuits is separate and does not affect the amplifying properties. The functions of a separate power amplifier and signal attenuator are also left.
Ok. Thank all of you for your patience with me.
I don't know how many times I have promised to address the different challenges in upgrading a Double Differential Topology Rotel (eg. the RA-970/980/990 series) from that of a traditional three-stage Rotel as eg. the RA-800/900/01,02,etc. series), etc. which I have already covered in this thread.
I am planning to break down the many issues into a number of separate posts to make it perhaps more readable.
If at any point some of you would want me to elaborate on certain things, please do ask away. Note that I am not any form of an expert on - well, anything, but I have had a deep physical dive into how to upgrade these Rotels using "Double Differential Input Push-Pull EF-VAS Topology" - DDiPPEFVAS (for short).
So, here goes.....
Double Differential Input Push-Pull EF-VAS Bias Revisited
The traditional single LTP / single VAS amplifier topology uses one current source for the LTP and another for the VAS / bias standing current. This easily defines and stabilises the DC operation independently in each stage.
Replacing the LTP collector resistor R1 with a current mirror is therefore relatively straightforward. The actual value of the V(?) voltage is determined by the VAS current from CS2 through R2 + one Vbe - and the circuit is quite forgiving and it works excellently in most cases.
Double-Differential input circuit topologies with push-pull VAS on the other hand, uses current sources for the two LTP’s, but none for the VAS stage. So, here the voltage across the LTP collector resistors now forward controls the VAS current.
But when you want to put in current mirrors for the input LTP’s instead of the traditional LTP collector resistors, controlling the current through VAS pairs and the bias transistor becomes tricky, if not impossible........
I don't know how many times I have promised to address the different challenges in upgrading a Double Differential Topology Rotel (eg. the RA-970/980/990 series) from that of a traditional three-stage Rotel as eg. the RA-800/900/01,02,etc. series), etc. which I have already covered in this thread.
I am planning to break down the many issues into a number of separate posts to make it perhaps more readable.
If at any point some of you would want me to elaborate on certain things, please do ask away. Note that I am not any form of an expert on - well, anything, but I have had a deep physical dive into how to upgrade these Rotels using "Double Differential Input Push-Pull EF-VAS Topology" - DDiPPEFVAS (for short).
So, here goes.....
Double Differential Input Push-Pull EF-VAS Bias Revisited
The traditional single LTP / single VAS amplifier topology uses one current source for the LTP and another for the VAS / bias standing current. This easily defines and stabilises the DC operation independently in each stage.
Replacing the LTP collector resistor R1 with a current mirror is therefore relatively straightforward. The actual value of the V(?) voltage is determined by the VAS current from CS2 through R2 + one Vbe - and the circuit is quite forgiving and it works excellently in most cases.
Double-Differential input circuit topologies with push-pull VAS on the other hand, uses current sources for the two LTP’s, but none for the VAS stage. So, here the voltage across the LTP collector resistors now forward controls the VAS current.
But when you want to put in current mirrors for the input LTP’s instead of the traditional LTP collector resistors, controlling the current through VAS pairs and the bias transistor becomes tricky, if not impossible........
hi thank you so much for sharing your very remarkable knowledge and valuable experience
i have two questions
1) do you deem the double differential topology superior on principle to the single one?
2) i tend to like better simpler circuits i. e. with lower parts count because they should be easier to optimize. This is something i rarely see Designers seem to like more complex circuits with the notable exception of Mr Pass designs . what is your opinion?
Thanks a lot and kind regards gino
i have two questions
1) do you deem the double differential topology superior on principle to the single one?
2) i tend to like better simpler circuits i. e. with lower parts count because they should be easier to optimize. This is something i rarely see Designers seem to like more complex circuits with the notable exception of Mr Pass designs . what is your opinion?
Thanks a lot and kind regards gino
HI Gino,
1) No.
The DDiPPVAS looks very nice on the schematic with everything beautifully balanced and mirrored etc, but getting excellent performance is (among other things) relying on finding transistor pairs of closely matching properties - particularly between NPN and PNP devices.
And the specs that are obtainable from both these topologies are very similar. Any audible differences? Hmm.. maybe, but I haven't done any blind A/B test.
2) I don't think it is that simple.
Yes, I also prefer low component count designs (Hiraga's 'Le Monstre' is a fine example), but it is a class A amp and so I believe are most of Pass' low component designs(?). In the lower idle power class AB/B there are a number of non-linear issues that need more added components to straighten out.
Per
1) No.
The DDiPPVAS looks very nice on the schematic with everything beautifully balanced and mirrored etc, but getting excellent performance is (among other things) relying on finding transistor pairs of closely matching properties - particularly between NPN and PNP devices.
And the specs that are obtainable from both these topologies are very similar. Any audible differences? Hmm.. maybe, but I haven't done any blind A/B test.
2) I don't think it is that simple.
Yes, I also prefer low component count designs (Hiraga's 'Le Monstre' is a fine example), but it is a class A amp and so I believe are most of Pass' low component designs(?). In the lower idle power class AB/B there are a number of non-linear issues that need more added components to straighten out.
Per
One problem with low parts count designs is that the actual active parts in the signal path are not as linear. That is the entire point of the more complicated designs. I would say that simpler is certainly not better from a performance point of view.
A more simple design is highly attractive to manufacturers for a few reasons. It is less expensive in parts and less complicated in PCB design, plus there are fewer parts that can fail (not really a big concern with signal transistors). Then you have some professional designers out there that actually are not that good at their job, and these designs challenge them. In defense of their choices, many designers / manufacturers will claim "that simpler is better". The more complicated voltage amplifier stages are in fact more linear, have higher transconductance and have far better noise immunity and better CMRR. These are all strongly positive traits.
Of course you can improve the performance of a less complicated or well built amplifier or design, but you will be limited by those other problems. So the more complicated voltage amplifier stages only really pay full dividends when the rest of the amplifier design is up to snuff.
There is such a thing as too complicated of course, some designs through the early 80's and 90's were just simply over-complicated without a real purpose. These caused a lot of trouble on the bench when things went wrong. But you can't use those as an excuse to say properly executed designs are over-complicated because the performance is really improved in measurable and audible ways. An oversimplification with a statement like (the fewer parts, the better" generally indicate a lack of understanding on the subject.
-Chris
A more simple design is highly attractive to manufacturers for a few reasons. It is less expensive in parts and less complicated in PCB design, plus there are fewer parts that can fail (not really a big concern with signal transistors). Then you have some professional designers out there that actually are not that good at their job, and these designs challenge them. In defense of their choices, many designers / manufacturers will claim "that simpler is better". The more complicated voltage amplifier stages are in fact more linear, have higher transconductance and have far better noise immunity and better CMRR. These are all strongly positive traits.
Of course you can improve the performance of a less complicated or well built amplifier or design, but you will be limited by those other problems. So the more complicated voltage amplifier stages only really pay full dividends when the rest of the amplifier design is up to snuff.
There is such a thing as too complicated of course, some designs through the early 80's and 90's were just simply over-complicated without a real purpose. These caused a lot of trouble on the bench when things went wrong. But you can't use those as an excuse to say properly executed designs are over-complicated because the performance is really improved in measurable and audible ways. An oversimplification with a statement like (the fewer parts, the better" generally indicate a lack of understanding on the subject.
-Chris
Double Differential Input Push-Pull EF-VAS Bias Revisited
Let’s first look at the DDiPPVAS circuit.
At a first glance on the schematic, it looks like someone has just thrown a handful of transistors into it and it can seem a bit chaotic, at least to the untrained eye.
So I have tried to redraw and perhaps clarify what is going on here.
I hope that the colouring shows that there are actually two identical "single stage" input LTP circuits (although mirrored PNP and a NPN). Each handles either the positive or negative part of the input signal and feeds it to two mirrored VAS "push-pull" transistors before it continues to the power stage, dc separated by the Bias transistor circuit.
When traditional LTP collector resistors (Ri) are used, the LTP current sources’ 2mA’s (or rather half of it) determines the base voltage Vb of each VAS stage, and thereby indirectly each VAS collector current through its emitter resistor Re. This works, but even so, the two VAS's have to eventually agree on a shared current value and any difference in opinion leads to an output offset. Further, the thermal stability of the VAS/bias current is limited – just try it.
After the initial turn on of such an amplifier (e.g. RA-971), the bias current quickly increases from about half to more than double the desired set value, and it takes about 10-15 minutes to stabilise back down as the bias transistor on the heat sink eventually heats up and gets to temperature in a slow, lagging thermal compensation.
Forward control by non-low impedance input circuits of higher (and potentially destructive) power stage currents is...well....really not an optimal engineering design.
And if you then want to put in LTP current mirrors, things get much worse as the VAS and bias current is no longer controlled at all.......
TBC,
Per
Let’s first look at the DDiPPVAS circuit.
At a first glance on the schematic, it looks like someone has just thrown a handful of transistors into it and it can seem a bit chaotic, at least to the untrained eye.
So I have tried to redraw and perhaps clarify what is going on here.
I hope that the colouring shows that there are actually two identical "single stage" input LTP circuits (although mirrored PNP and a NPN). Each handles either the positive or negative part of the input signal and feeds it to two mirrored VAS "push-pull" transistors before it continues to the power stage, dc separated by the Bias transistor circuit.
When traditional LTP collector resistors (Ri) are used, the LTP current sources’ 2mA’s (or rather half of it) determines the base voltage Vb of each VAS stage, and thereby indirectly each VAS collector current through its emitter resistor Re. This works, but even so, the two VAS's have to eventually agree on a shared current value and any difference in opinion leads to an output offset. Further, the thermal stability of the VAS/bias current is limited – just try it.
After the initial turn on of such an amplifier (e.g. RA-971), the bias current quickly increases from about half to more than double the desired set value, and it takes about 10-15 minutes to stabilise back down as the bias transistor on the heat sink eventually heats up and gets to temperature in a slow, lagging thermal compensation.
Forward control by non-low impedance input circuits of higher (and potentially destructive) power stage currents is...well....really not an optimal engineering design.
And if you then want to put in LTP current mirrors, things get much worse as the VAS and bias current is no longer controlled at all.......
TBC,
Per
Last edited:
Hi there Per,
Now I "understand" why this subject takes so long so the extra effort you put in to make it more understandable for those who don't have the knowledge is much appreciated, thank you so much.
And even though my knowledge of amplifier architecture doesn't come close within light years to yours, your drawings in red and blue are very unveiling for me.
I cannot wait to see where this is getting to.
About the simpler circuitry/less component designs, what about Susan Parkers Zeus amplifier?
Now I "understand" why this subject takes so long so the extra effort you put in to make it more understandable for those who don't have the knowledge is much appreciated, thank you so much.
And even though my knowledge of amplifier architecture doesn't come close within light years to yours, your drawings in red and blue are very unveiling for me.
I cannot wait to see where this is getting to.
About the simpler circuitry/less component designs, what about Susan Parkers Zeus amplifier?
Hi Paulus,
The Zeus....Well, transformers were/are a necessity in high voltage tube/valve amps. They are not exactly super linear or low distortion - particularly even harmonics, but they definitely have a dedicated following. And I admit, I too like to listen to a good, clean vinyl record played over a tube amp now and then.
But if you look at the spec graph, the THD increases from 0.01 to 1% over 25W power output - much like tube amps. And the 'General Circuit' does not show more than the basic principle. But hey, I have never actually listened to a Zeus amp, so I am in no position to criticize.
However, for guitar - tube amps (and their transformers) reign supreme IMHO.
In my younger days when I played in bands I could only afford a cheap Tokyo Sound Combo valve amp. It sounded absolutely terrible, so being an electronics uni student, I of course went straight in and upgraded its guts with some circuit and component changes, MESA Boogie tubes and a proper Celestion speaker.
That improved the amp - but not my playing.
Cheers,
Per
The Zeus....Well, transformers were/are a necessity in high voltage tube/valve amps. They are not exactly super linear or low distortion - particularly even harmonics, but they definitely have a dedicated following. And I admit, I too like to listen to a good, clean vinyl record played over a tube amp now and then.
But if you look at the spec graph, the THD increases from 0.01 to 1% over 25W power output - much like tube amps. And the 'General Circuit' does not show more than the basic principle. But hey, I have never actually listened to a Zeus amp, so I am in no position to criticize.
However, for guitar - tube amps (and their transformers) reign supreme IMHO.
In my younger days when I played in bands I could only afford a cheap Tokyo Sound Combo valve amp. It sounded absolutely terrible, so being an electronics uni student, I of course went straight in and upgraded its guts with some circuit and component changes, MESA Boogie tubes and a proper Celestion speaker.
That improved the amp - but not my playing.
Cheers,
Per
Last edited: