Hi Thorsten,
Sorry for being late coming back to this.
On average, people trained in "the art", or whatever you want to consider will produce the desired results. Trouble begins where some people assume they are qualified to take action go ahead and take whatever action they have dreamed up as being appropriate. To focus this comment back to audio, for over 35 years I have had to clean up after would-be electronics technicians. It is more difficult to clean up after self proclaimed audio engineers though. Their actions reach much further in society. Maybe you are referring to these people? Those who claim to be proficient in a skill that they are not. You generally can not tell them they don;t measure up either. It is very interesting that as ability decreases, pride and a swollen head increases. Keep in mind I am not pointing at anyone here, nor inferring any of our present company might fit the description I gave.
There are often times when an individual who has taken the time to study and learn can in fact do a good job. However, the averages do not support this.
Hi cliffforest,
I think you are pointing out that just 'cause you have the tools does not guarantee your answers will be correct.
However -
If you do not possess the tools, you ain't got any answers at all. Not even a ballpark type answer! That's the worst kind of being in the dark I think.
If you look into technical seminars on the newer classes of digital instruments, you will see that test equipment manufacturers are taking pains to address this very issue. With the cost of these instruments, who can afford not to become educated in their proper use and limitations??
Your average professional does update their knowledge regularly, specifically with regard to any new equipment they will be using.
I trust real professional engineers (and doctors ...) to a point. Once they show me they are part of the majority that is a professional, they have my trust. That does not mean I would take what I'm told blindly, but I sure as heck am not going to debate it with them!
For the record, I don't know how to use a simulator. I just use a calculator and build the darn thing. Usually it works, but needs some component values adjusted. No biggie. One of these days I'll learn how to do this.
-Chris
Sorry for being late coming back to this.
Ah come on now! Get real! Now you are using a minority to discredit the majority! Not only that, but it's usually the contractor and suppliers that make building fall down, not the poor sod who designed and spec'd a building.
On average, people trained in "the art", or whatever you want to consider will produce the desired results. Trouble begins where some people assume they are qualified to take action go ahead and take whatever action they have dreamed up as being appropriate. To focus this comment back to audio, for over 35 years I have had to clean up after would-be electronics technicians. It is more difficult to clean up after self proclaimed audio engineers though. Their actions reach much further in society. Maybe you are referring to these people? Those who claim to be proficient in a skill that they are not. You generally can not tell them they don;t measure up either. It is very interesting that as ability decreases, pride and a swollen head increases. Keep in mind I am not pointing at anyone here, nor inferring any of our present company might fit the description I gave.
There are often times when an individual who has taken the time to study and learn can in fact do a good job. However, the averages do not support this.
I'm trying to understand what you mean here. I hope you aren't digging up the old subjective vs objective argument again. To lay matters at rest, anyone who is designing audio equipment is well advised to both use the best instrumentation they can get - and to listen to the thing once they have a prototype up and running. The chances of creating something really good are greatly diminished without listening or without measuring what the circuit is doing. Those audio "engineers" who make a mess of things are those who do not use test equipment. The person who uses good engineering practices but doesn't listen will create something that at least doesn't burn up, although it may have an underwhelming sound quality. There are many examples of each type of failure. You need a bit of both approaches to do good work, period.
I think we both agree on this. That's why I was confused by the earlier statement you posted (quoted).
Hi cliffforest,
I think you are pointing out that just 'cause you have the tools does not guarantee your answers will be correct.
However -
If you do not possess the tools, you ain't got any answers at all. Not even a ballpark type answer! That's the worst kind of being in the dark I think.
Well, no. With any luck, good quality tools were purchased and the user is properly educated in their use (and accuracy!). Some digital equipment displays are only accurate to 2 digits out of 3 1/2 or 4 3/4. Knowing how a meter responds to various signals is of course important, but you have to have a look at what you are measuring (use a 'scope) to see what the meter is looking at. Digital 'scopes are another item that needs understanding. Having an old (even 20 MHz) analog 'scope around is always good insurance.
If you look into technical seminars on the newer classes of digital instruments, you will see that test equipment manufacturers are taking pains to address this very issue. With the cost of these instruments, who can afford not to become educated in their proper use and limitations??
Your average professional does update their knowledge regularly, specifically with regard to any new equipment they will be using.
I trust real professional engineers (and doctors ...) to a point. Once they show me they are part of the majority that is a professional, they have my trust. That does not mean I would take what I'm told blindly, but I sure as heck am not going to debate it with them!
For the record, I don't know how to use a simulator. I just use a calculator and build the darn thing. Usually it works, but needs some component values adjusted. No biggie. One of these days I'll learn how to do this.
-Chris
Last edited:
But how many people remember how to use a Simpson 260 and compensate for the low input impedance?
On the DH-120: Out of curiousity I went ahead and did the best I could manage with a simulation, although lacking models for those DMOS parts I used comparable capacitance/current/power vertical ones. But these have much higher magnitude gate-source turnon voltages, so I had to drastically change the values in the "Vbe multiplier" to get the output devices comparably biased, which pieced together from the manual's tables of voltages and the specified DMOS parts is somewhere in the vicinity of 250mA at quiescence. This amp must run quite warm.
One of the advantages of the specified parts exploited by the design is the operation of them at, approximately, a drain-current zero-temperature-coefficient point. As has been remarked before, a similar operating point for vertical MOS usually occurs at impractically large currents and power dissipations.
Anyway, subject to the plethora of caveats about neglected variables in sim, the distortion performance from simulation seems in line with the stated performance of the real amplifier. One of the somewhat unusual features appears to be a THD relatively invariant with signal frequency, rising only slightly between 1kHz and 10kHz.
I may do some more work to see what the open-loop phase/gain looks like.
Brad
One of the advantages of the specified parts exploited by the design is the operation of them at, approximately, a drain-current zero-temperature-coefficient point. As has been remarked before, a similar operating point for vertical MOS usually occurs at impractically large currents and power dissipations.
Anyway, subject to the plethora of caveats about neglected variables in sim, the distortion performance from simulation seems in line with the stated performance of the real amplifier. One of the somewhat unusual features appears to be a THD relatively invariant with signal frequency, rising only slightly between 1kHz and 10kHz.
I may do some more work to see what the open-loop phase/gain looks like.
Brad
I've not seen any frequency-dependent difficulties with Spice filter analysis --- perhaps you could provide an example. Although I agree that with relatively simple topologies (Sallen-Key for example) they are easy to design without simulation.
As for CG0/NP0, other than poor volumetric efficiency, and high cost for large values, most have seemed to be quite adequate. I remember reading about one made for the International Space Station. IIRC, 270uF (yes microfarads!), very low inductance, one big mother, about the size of a small loaf of bread. Don't ask the price. It was for a key switching regulator.
I specified an NP0 recently for a non-audio product because of the requirement for a low temperature coefficient of capacitance. The largest one I could readily find in surface mount was 22nF, which worked out to be about right for the application. Unfortunately the tempco of the resonant inductor's loss turned out to be quite peculiar, but I will resist further off-topic excursions
Brad
Brad,
Even with a simple Sallen-Key filter I found spice to not always work to the right degree of precision, but it depends on what I am doing. To give two examples where it went wrong with my last project. The first is a S-K high pass with a highish Q, somewhere around 1.2. Objective is a subsonic filter for a BR-enclosure. High Q gives you a nice fast roll off below the BR tuning frequency, where the enclosure will no longer produce any usefull bass anyways. If the peak of the filter hits exactly at the BR tuning frequency, you can really hit the woofer hard because cone movements are very small at this point. Even when I fed the sim with exact measurements of the components, the sim result would be off from what the physical build with the same components produced (as in: SIM says 50, build does 60Hz). The second example of things that tend to go wrong in a sim are compound high/low shelf filters to compensate for driver characteristics. I might be doing something wrong, but I tried hard to make the sims work, because that would save me quite a bit of time.
On NGO: I used to order from RS till they started to get too many orders wrong, so I changed to Mouser just to see how that would work out. RS only has NGO up till 22 nF (tth), but Mouser had them up to 100 nF. Not really small in size, but a good deal smaller than the NOS polystyreen I have been using. I just don't understand the high level of pricing, probably some patent underlaying the technology.
vac
Last edited:
Hi,
The "majority" tends to be those who do not have a lot of challenge. A good friend of mine is a retired architect (and used to teach) and I used to work for the NHS in the UK on financial systems, so I know precisely the bill for medical negligence (the stuff that actually gets paid is the tip of the iceberg BTW) for many years running for North London.
I would not that that those that get caught by the consequences are the majority, but "sloppy work" is by far more common than you would make us believe, it is just that usually people are forced to allow for very generous safety margins, so sloppy work does not collapse too many building and sloppy doctors do not maim too many payshunts...
Sure.
They probably all used fake bricks imported from China.
All of them.
That depends on what you define as "desired results". I have reason to consider this untrue at least as far as my personally desired results and experience go.
This I agree with. As far as many supposed "professionals" who took the time to study and learn (or more like to drink) are concerned.
Am I really that obtuse?
True science uses controlled experiments, good experimental design etc. et al ad nausaeam... I have incurred RSI just typing about this sh...tuff for the last decade plus.
Do you really need me to point out the number of times in this thread where individuals have asserted that a certain measurement was a measure of quality (if only by the reverse, by claiming that comparably poor performance in this measurement meant an automatic failure) without providing the least shred of evidence to support their assertions?
AND more crucially with no such evidence as to the usefulness of these measurements posited in human knowledge, but often the reverse in fact having been posited?
Is that any better than to claim that no measurements matter and all we ever need to do is listen?
That is simply unscientific hogwash and in fact pure "belief" or "faith", not backed by any evidence on either side.
What I find amusing is that again and again eventually measurement methods are found that seem to quantify what the "listeners" have long claimed to hear, from differences in capacitors to (wait for it)
...
...
...
...
even capacitor break-in....
I will not disagree that the other side in the argument is as guilty, but that does not allows for an excuse or make the first right. The shoe fits both ways.
Though I guess to me it is:
"The say two wrongs won't make it right,
but it's suiting me just fine"
Kandi - Don't Think I'm Not - YouTube
In my view we all would be better off with fewer challenges and confrontation and with instead shouting at each from deep trenches over a no-man's land, with just sitting down over a nice cup of tea and cheese crumpets and talk it over and either find common ground or to agree to disagree. Trust me, on my side it is not for lack of trying.
As someone who has taken stats at uni twice with a few years inbetween and two radically different eduction systems, I can assure you that hell will freeze over, the universe will die a heat death (in other words we will see a "Gnab Gib") and trees will grow with their leafs in the ground and their roots in the air several times over before that actually happens.
And yes, I made that statistic up on the spot, which makes it about as true as most of them...
Ciao T
The "majority" tends to be those who do not have a lot of challenge. A good friend of mine is a retired architect (and used to teach) and I used to work for the NHS in the UK on financial systems, so I know precisely the bill for medical negligence (the stuff that actually gets paid is the tip of the iceberg BTW) for many years running for North London.
I would not that that those that get caught by the consequences are the majority, but "sloppy work" is by far more common than you would make us believe, it is just that usually people are forced to allow for very generous safety margins, so sloppy work does not collapse too many building and sloppy doctors do not maim too many payshunts...
Sure.
They probably all used fake bricks imported from China.
All of them.
That depends on what you define as "desired results". I have reason to consider this untrue at least as far as my personally desired results and experience go.
This I agree with. As far as many supposed "professionals" who took the time to study and learn (or more like to drink) are concerned.
Am I really that obtuse?
True science uses controlled experiments, good experimental design etc. et al ad nausaeam... I have incurred RSI just typing about this sh...tuff for the last decade plus.
Do you really need me to point out the number of times in this thread where individuals have asserted that a certain measurement was a measure of quality (if only by the reverse, by claiming that comparably poor performance in this measurement meant an automatic failure) without providing the least shred of evidence to support their assertions?
AND more crucially with no such evidence as to the usefulness of these measurements posited in human knowledge, but often the reverse in fact having been posited?
Is that any better than to claim that no measurements matter and all we ever need to do is listen?
That is simply unscientific hogwash and in fact pure "belief" or "faith", not backed by any evidence on either side.
What I find amusing is that again and again eventually measurement methods are found that seem to quantify what the "listeners" have long claimed to hear, from differences in capacitors to (wait for it)
...
...
...
...
even capacitor break-in....
I will not disagree that the other side in the argument is as guilty, but that does not allows for an excuse or make the first right. The shoe fits both ways.
Though I guess to me it is:
"The say two wrongs won't make it right,
but it's suiting me just fine"
Kandi - Don't Think I'm Not - YouTube
In my view we all would be better off with fewer challenges and confrontation and with instead shouting at each from deep trenches over a no-man's land, with just sitting down over a nice cup of tea and cheese crumpets and talk it over and either find common ground or to agree to disagree. Trust me, on my side it is not for lack of trying.
As someone who has taken stats at uni twice with a few years inbetween and two radically different eduction systems, I can assure you that hell will freeze over, the universe will die a heat death (in other words we will see a "Gnab Gib") and trees will grow with their leafs in the ground and their roots in the air several times over before that actually happens.
And yes, I made that statistic up on the spot, which makes it about as true as most of them...
Ciao T
Hi,
The original under 15W/Channel. Not sure that counts as "warm".
ALL of these observations are explained by:
1) Reasonable amounts of degeneration (that accounts for a close match between sim and real)
2) The > 20KHz open loop bandwidth (that accounts for the lack in significant rise in THD with frequency)
Ciao T
The original under 15W/Channel. Not sure that counts as "warm".
ALL of these observations are explained by:
1) Reasonable amounts of degeneration (that accounts for a close match between sim and real)
2) The > 20KHz open loop bandwidth (that accounts for the lack in significant rise in THD with frequency)
Ciao T
Agree. And as far as bias, this was based on just the low-resolution owner's manual voltage measurements and the curves on the Hitachi datasheet for the N part, which seems to coincide with the zero tempco point for the device. Of course the curves are simply meant to be typical, not guaranteed.
I did do an open-loop sim and the numbers are an open-loop midband gain of 83.9dB (with an 8 ohm resistive load) -3dBr at about 22kHz. The open-loop distortion for 20W out is about 1.6% at 10kHz (harmonics to 50kHz included). Phase margin is pretty decent, diminishing to 45 degrees at about 160kHz.
So, nice job Erno!
By the way, there is an error in the table of voltages, with different entries for the emitters of the two current mirror devices. But if the measurement for the emitter of Q6 is correct, only one of the different ones can be correct, the one more positive
And again placing a lot of faith in the numbers being exact, this implies that the second stage current is running around 3.6mA, ageeing well with the "Balance" pot being roughly at the midpoint. I wouldn't want to find out what happens if it's adjusted to its minimum value.
I suspect they used a Simpson for the voltage measurements as the same node has different values by a bit. Bias is set to 125mV by the manual, but I noticed the pots were at max.
I get:
C5/R12 @ 3.4, but not how to calculate the zero
C8/R16 @ 5.9K but no clue on the"invisible" pole
No idea how one calculates C6 or the"invisible pole"
R13/C4 @ 1.5Hz
C3/R11 @ 22K which seems a shade low.
I noticed it the EB 60 and 100's he did like the pole on the output of the VAS.
The gate resistors were all messed up. With the new Exicons, I am setting them per app note at 1M, which is 680 on Q201, 470 on Q 202 and moving the resistors out to right at the gate pin. I am sure that is going to mess something up in there as the original parts were set asymmetrically at 250K and 1.2M.
My rookie attempt at a Bode plot came out at -270, so clearly I don't have a clue yet.
Don't I want sufficient rise times et, al so the poles and zeros are above the usable BW by 2X or so? If so a cap across Q8 and increasing R11 would be good things? ( adjust C3 obviously.)
If I lift C2 and feed it some square waves, then I should be able to play with these parts and see what damage I am causing. I wonder what renting a current probe costs.
P2 at zero, 65 ma for a little while. It would be safer to increase R7 to 100 I would think.
Measuring the parts I took out, Q 3/4 and 1/2 were not very well matched, so I doubt that 1.6% @ 10K OL figure was actually attained. I bought enough new ones I hope to do better than they did.
Enough computer time, need to measure some transistors and power it back up.
I get:
C5/R12 @ 3.4, but not how to calculate the zero
C8/R16 @ 5.9K but no clue on the"invisible" pole
No idea how one calculates C6 or the"invisible pole"
R13/C4 @ 1.5Hz
C3/R11 @ 22K which seems a shade low.
I noticed it the EB 60 and 100's he did like the pole on the output of the VAS.
The gate resistors were all messed up. With the new Exicons, I am setting them per app note at 1M, which is 680 on Q201, 470 on Q 202 and moving the resistors out to right at the gate pin. I am sure that is going to mess something up in there as the original parts were set asymmetrically at 250K and 1.2M.
My rookie attempt at a Bode plot came out at -270, so clearly I don't have a clue yet.
Don't I want sufficient rise times et, al so the poles and zeros are above the usable BW by 2X or so? If so a cap across Q8 and increasing R11 would be good things? ( adjust C3 obviously.)
If I lift C2 and feed it some square waves, then I should be able to play with these parts and see what damage I am causing. I wonder what renting a current probe costs.
P2 at zero, 65 ma for a little while. It would be safer to increase R7 to 100 I would think.
Measuring the parts I took out, Q 3/4 and 1/2 were not very well matched, so I doubt that 1.6% @ 10K OL figure was actually attained. I bought enough new ones I hope to do better than they did.
Enough computer time, need to measure some transistors and power it back up.
I get the feeling that I'm looking at a different manual. The only thing I saw about setting bias when I just looked for it was using an ammeter in place of one of the fuses, and adjusting for 150mA.
As Thorsten mentions, degeneration with passives helps alleviate mismatch issues, so R3 and R4 make the input device matching a bit less important. The current mirrors are a bit more sensitive as there is no emitter degeneration. But there is also base current loss and differences in characteristics with different collector voltages.
Looking for the dominant distortion mechanisms though, I'd guess the second stage is the bigger contributor. Not sure what the open loop distortion of the DMOS with loading is though, so maybe that is even more significant. They are not all that high-transconductance.
The different bias measurements I referred to are the emitters of Q3 and Q4, which are connected according to the schematic by a trace, so it's difficult to develop much of a voltage diferential
But maybe the schematic is wrong.The envelope please. Again my FET models are a lousy approximation to the actual 120 parts, but driven from a voltage source and driving 8 ohms the output stage distortion with substitute FETs and 20W into 8 is about 1.3% (at 1kHz, but it's not terribly freq-dependent in the audio band). So that's probably the most significant for open loop amp distortion. It's also mostly even-order, so if the Hitachi parts are better complements that will come down.
Are you driving the filter input with the same impedance as used for simulation? S-K is usually not too sensitive, the IGMF (infinite gain multiple feedback) structures are a lot more so, especially if one is trying to do a three-pole. That latter has bitten me before, when I was trying to come up with complementary polarity signal chains with the same filter response but opposing polarities (and didn't want to throw down an additional inverting stage).
Last edited:
The invisible poles, are they the poles from the Cbe ? So I would have two more from Q10/Q11 and the final one I set on the outputs be adjusting the source resistors.
Think I need to do a spreadsheet to make the plots. The ONE thing I hold against Excel is they never have provided log/log graphs.
Think I need to do a spreadsheet to make the plots. The ONE thing I hold against Excel is they never have provided log/log graphs.
Hi Thorsten,
In housing these days, it seems impossible to find floors that are level and even. Of those that are now, they'll be creaking and dipping in a few years. Blows me away. My house now has floors that creak ... a lot. I'm not happy about it either.
What can I say?
As for the next comments you made, I am in agreement with you. I'm pretty much on the record as to what my views are on what makes a successful designer.
Tea sounds fine, when?
-Chris
I agree about people too involved in arguing while missing the point of it all. Nicely put by the way ...
I have a feeling that you are a fairly picky person about how to do things, and that is perfectly fine. I suffer the same issues in that regard. I had only three technicians work for me who I considered to be good in a 16 year period. I'm sure these fellows were exceptional, and I miss each one of them. Most techs that I kept were okay. They did the work properly most of the time and were perhaps not as clean and didn't make adjustments just so, but they did much better work than I saw from other places. If I were to adjust my levels downwards a bit, then these people were very good (making the other three gods of the bench). A little lower quality brings you to the most common people I saw in the industry. I let them go eventually, or maybe they let me go? Either way, the standards I kept made life difficult for them so that we were not a good fit. The next lower grade of technicians were those who either didn't pass the interview, or they didn't survive the week on the bench. I'm not going to talk about "the others". That was something I didn't understand at all. Someone who doesn't care about their workmanship. They are wasting their lives.
Actually, no. I came out of retirement to service audio equipment again because of exactly this issue. Old customers actually searched and found me to ask if I would fix their - whatever. Now that is sad. These days it seems that the bar has been set pretty low in the service industry. However, it is also perfectly understandable when you look at the service rates and lack of technical support from distributors. I am not surprised, and the industry has done this to itself and their customers.
No. It's more like using cement and mortar that isn't up to spec. The bricks are simply those things that fall because they are not fixed into place very well.
In housing these days, it seems impossible to find floors that are level and even. Of those that are now, they'll be creaking and dipping in a few years. Blows me away. My house now has floors that creak ... a lot. I'm not happy about it either.
It is too easy to agree with you. Having been in this industry for far too long, I have seen quality of everything from design to build and service materials drop over time. I watch as really good techs hang up their iron and move to industry (as I did). Who suffers the most then? Well, the customers do, no doubt. The other are those good techs who still try to do good work. They are the ones left holding the bag because to fix the issues takes a long time. They can not charge the customer for all their time. the families of these people also do without as a result.
No, we each look at things a little differently. Since I wasn't sure what you meant, I commented hoping you would confirm or explain so I would be more sure.
What can I say?That happens all the time in most threads. Maybe those who are least sure of themselves are the most prickly. Can't let their pride get bruised!
As for the next comments you made, I am in agreement with you. I'm pretty much on the record as to what my views are on what makes a successful designer.
Was it fun? The difference is, you weren't trying to prove a point by misleading anyone. The point was made and understood.
Tea sounds fine, when?
-Chris
Hi,
Yes. Very. This 100MHz GBWP!
This amp is still, as Mr. Davis once put it "Miles Ahead" of many today.
What Sim are you using? If LT-Spice you can find files with decent lateral models in a number of threads.
It would be fin to sim the changes to better small signal transistors and a dual die (read two parallel fets) output part...
Ciao T
Yes. Very. This 100MHz GBWP!
This amp is still, as Mr. Davis once put it "Miles Ahead" of many today.
What Sim are you using? If LT-Spice you can find files with decent lateral models in a number of threads.
It would be fin to sim the changes to better small signal transistors and a dual die (read two parallel fets) output part...
Ciao T
Hi,
No, from Re. In theory we need to add up even more poles and zeros as we also have the transistors Cob...
Try LT-Spice to stand a chance.
Seeing all what has been discussed, I would suggest the following:
1) Use dual die parts as outputs at double the Iq, this way you reduce distortion and make the amplifier more load tolerant, the one are where the original DH-120 is a bit iffy.
2) Replace all the signal transistors with the Pair's Wahab suggested, in this kind of Amp (but not necessarily in my kind of Amp) these will do an excellent Job.
Make Q1, Q2 and Q5 a 2SC1775A (highest beta class)
Make Q3, Q4 and Q6 a 2SA872A (highest beta class)
Make Q9, Q11 a 2SC1845 (highest beta class)
Make Q7, Q10 a 2SA992 (highest beta class)
Upgrade the cap's as discussed before. Upgrade the PSU as much as possible.
The result should be a very decent power-amp that probably will be able to hold it's own against a lot of stuff out there.
Given the results EB got I would not really monkey about much, expect possibly degenerating the current mirror and VAS a bit...
Ciao T
No, from Re. In theory we need to add up even more poles and zeros as we also have the transistors Cob...
Try LT-Spice to stand a chance.
Seeing all what has been discussed, I would suggest the following:
1) Use dual die parts as outputs at double the Iq, this way you reduce distortion and make the amplifier more load tolerant, the one are where the original DH-120 is a bit iffy.
2) Replace all the signal transistors with the Pair's Wahab suggested, in this kind of Amp (but not necessarily in my kind of Amp) these will do an excellent Job.
Make Q1, Q2 and Q5 a 2SC1775A (highest beta class)
Make Q3, Q4 and Q6 a 2SA872A (highest beta class)
Make Q9, Q11 a 2SC1845 (highest beta class)
Make Q7, Q10 a 2SA992 (highest beta class)
Upgrade the cap's as discussed before. Upgrade the PSU as much as possible.
The result should be a very decent power-amp that probably will be able to hold it's own against a lot of stuff out there.
Given the results EB got I would not really monkey about much, expect possibly degenerating the current mirror and VAS a bit...
Ciao T
More study. Again thanks. Bit by bit it is making more sense, which of course it the goal. A better amp is a side benefit. There is quite a gap between reality and the text book. Both books claim if the dominant pole is low enough, 22K in this case, and the rest are sufficiently far enough away, it will be stable by definition. Clearly not true. I also do not see why the global feedback pole is higher than the BW of the output.
Starting to look at various spice distro's. The original freeware version of PSpice is limited to 10 transistors. Guess I may have to dust off an old box and fire up Ubuntu on it. This is going to take some time.
Starting to look at various spice distro's. The original freeware version of PSpice is limited to 10 transistors. Guess I may have to dust off an old box and fire up Ubuntu on it. This is going to take some time.
When I was using spreadsheets for modeling, I plotted the output in decibels, using "20*log(imabs(whatever))", assuming you're using complex arithmetic for the calculations.
TVR, which components are you referring to as forming the global feedback pole (C7 is for compensating the pole formed by the feedback resistor network and the input transistor capacitance) ?
Excel allows you to format either axis as log scale.
Thanks
-Antonio
The text books can answer for themselves, but 'low enough' for the dominant pole depends on the loop gain. The higher the mid-band loop gain, the lower the dominant pole has to be to be 'low enough'. Any decent textbook would make this clear. Either you are reading the wrong book, or you have misunderstood what you read.tvrgeek said:
Roughly speaking, the ratio between the dominant pole and the average of the next two up has to exceed the mid-band loop gain. A book might give you a more precise answer. I am just working from memory.
- Status
- Not open for further replies.