I am running an Aikido using four 6SN7s for a total current draw of around 23mA, or around 30mA if I add an extra 30K draw across the final cap of the PSU. Currently, my PSU is:
550VAC
5V4
0.78uF
10H choke
60uF
10H choke
60uF
B+ is around 317VDC at 24mA and I get a little 120Hz hum. PSUDII modelling indicates that I can reduce the ripple to sub mV levels by increasing the first cap to around 1.5uF. I can increase it further by increasing the final cap. I have a couple of options for this:
1. Parallel a 20uF with the final 60uF to give 80uF.
2. Parallel a 45uF with the final 60uF to give 105uF.
OR
1. Parallel a 20uF with the first 60uF to give 80uF.
AND
2. Parallel a 45uF with the second 60uF to give 105uF.
Anyway, is the level of capacitance too much? I expect that the answer is a very clear "No" but I'd like to be sure.
Thanks,
Charlie
550VAC
5V4
0.78uF
10H choke
60uF
10H choke
60uF
B+ is around 317VDC at 24mA and I get a little 120Hz hum. PSUDII modelling indicates that I can reduce the ripple to sub mV levels by increasing the first cap to around 1.5uF. I can increase it further by increasing the final cap. I have a couple of options for this:
1. Parallel a 20uF with the final 60uF to give 80uF.
2. Parallel a 45uF with the final 60uF to give 105uF.
OR
1. Parallel a 20uF with the first 60uF to give 80uF.
AND
2. Parallel a 45uF with the second 60uF to give 105uF.
Anyway, is the level of capacitance too much? I expect that the answer is a very clear "No" but I'd like to be sure.
Thanks,
Charlie
note - an old thread. just using the topic of an OP.
many times i thought about a sane usage of PSUD2. for sure it can be a handy tool to check some things here and there, but over the years i've seen it taken as an absolute authority. me included - i didn't know much about the PSU designs (and i'm still far from it) and i happily jumped on PSUD2 and fiddled with it until
a) there's no ringing/impact in the beginning - smooth, rounded start
b) continuous ripple as low as possible, desirably almost undetectable.
but .. this approach has led always to really big caps. or in reverse order - modelling known and very good amps almost everytime ended with a bad looking curve. you can make an excellent amp with two 40-50uf caps and lets say 10-15H choke or two. but if i construct these strategies in the program, i fail to make in nice there. always wants bigger chokes and most certainly 5 times the capacitance.
in other words, as i see it - PSUD2 can make an unexperienced person to solve problems that don't really exist, for example to add too much capacitance. it's just a program and can't calculate with SOUND. you just can't see it there, know the dynamics that capacitance alters, know how much ripple will you actually hear in the end etc.
all the best,
misok
many times i thought about a sane usage of PSUD2. for sure it can be a handy tool to check some things here and there, but over the years i've seen it taken as an absolute authority. me included - i didn't know much about the PSU designs (and i'm still far from it) and i happily jumped on PSUD2 and fiddled with it until
a) there's no ringing/impact in the beginning - smooth, rounded start
b) continuous ripple as low as possible, desirably almost undetectable.
but .. this approach has led always to really big caps. or in reverse order - modelling known and very good amps almost everytime ended with a bad looking curve. you can make an excellent amp with two 40-50uf caps and lets say 10-15H choke or two. but if i construct these strategies in the program, i fail to make in nice there. always wants bigger chokes and most certainly 5 times the capacitance.
in other words, as i see it - PSUD2 can make an unexperienced person to solve problems that don't really exist, for example to add too much capacitance. it's just a program and can't calculate with SOUND. you just can't see it there, know the dynamics that capacitance alters, know how much ripple will you actually hear in the end etc.
all the best,
misok
Newbies often obsess about ripple, so add huge caps and huge chokes to get rid of it. Meanwhile, their grounding methods inject charging pulses into the signal ground, they scatter a few ground loops around, and the rectifier current loop does magnetic induction all around. These latter problems are harder to simulate so go unnoticed at first.
Is the old empirical rue about 1000 uF or 2000 uF for 1 A is still good ?
In the same time when 4 000 uf is used to filter and charge a reference voltage as a zener diode in a local section on a a shunt reg, could the main cap staying at 1 000 uF when the final load only needs five less than 1 A !
To stay on the OP spirit I ask myself hox to deal with the smoothing cap which have to be important for avoiding pics and the second one aftyer the chocke as reservoir which must stay at minimum equal to the first smoothing one while not needing to be more than 100 uF for the load ?
In other words : how to deal between the time loading on a reservoir cap VS the smoothering needs ?
If the load doesn't need too much current and a cap is nearly full after 5 time period : does the first time period which load 50 % of a cap is not enough for a low need current even if the cap is huge ? (the load will use 5% of the load of the first Time period on this huge cap, so Ok ?)
Sorry if two cents, hope it's not Off topic as I try myself to understand how to filtering but staying on the maxi pics current needed by a load ! Because my very simple understanding being not a technician is the sound has more impact and is less dull when the cap is not to huge (not above its needs) !
In the same time when 4 000 uf is used to filter and charge a reference voltage as a zener diode in a local section on a a shunt reg, could the main cap staying at 1 000 uF when the final load only needs five less than 1 A !
To stay on the OP spirit I ask myself hox to deal with the smoothing cap which have to be important for avoiding pics and the second one aftyer the chocke as reservoir which must stay at minimum equal to the first smoothing one while not needing to be more than 100 uF for the load ?
In other words : how to deal between the time loading on a reservoir cap VS the smoothering needs ?
If the load doesn't need too much current and a cap is nearly full after 5 time period : does the first time period which load 50 % of a cap is not enough for a low need current even if the cap is huge ? (the load will use 5% of the load of the first Time period on this huge cap, so Ok ?)
Sorry if two cents, hope it's not Off topic as I try myself to understand how to filtering but staying on the maxi pics current needed by a load ! Because my very simple understanding being not a technician is the sound has more impact and is less dull when the cap is not to huge (not above its needs) !
Thanks DF96
did you mean than with a pic load of 1 A from the source (saying for example that its needs is 15V with an average 200 mA) a 1 000 or 2 000 smooting/reservoir (if no Pi filtering) is not enough (in the case of many pics at 1 A ?) ? Do you talk in relation to the particular OP needs or in general here ?
Or too much if the average needs is always 200 mA from the load ?
PSUDII can simulate this ?
I'm lost between the voltage smoothing and the current needs when it comes to find the cap size of a PS. The traffos I use are always big for my needs so rarely too weak, but I'm more lost about those question of timings in relation to the SQ/cap size, especially Q factor (bass, impacts...following difficult tempo...)
did you mean than with a pic load of 1 A from the source (saying for example that its needs is 15V with an average 200 mA) a 1 000 or 2 000 smooting/reservoir (if no Pi filtering) is not enough (in the case of many pics at 1 A ?) ? Do you talk in relation to the particular OP needs or in general here ?
Or too much if the average needs is always 200 mA from the load ?
PSUDII can simulate this ?
I'm lost between the voltage smoothing and the current needs when it comes to find the cap size of a PS. The traffos I use are always big for my needs so rarely too weak, but I'm more lost about those question of timings in relation to the SQ/cap size, especially Q factor (bass, impacts...following difficult tempo...)
Last edited:
Yes, and I'll explain why:
I consider the guide value of 1mF to 2.2mF per ampere of current to be a good guide, for low grade and medium grade continuous DC PSUs.
As a first step towards a target, it gets one going. For a high grade PSU, I double the capacitance again to 4m7F/A. Call them 1m, 2m & 5m, if you wish. I tend to adopt the high grade 4m7F/A for all line level amp/pre-amp/crossover stage PSUs, it costs nothing extra, (unless one adopts Sikorel caps). I have not designed low level stages, for example microphone/MM/MC amps. An ultra high grade guide figure should maybe adopted for these high gain stages.
Take a ClassAB amplifier running on +-50Vdc, drawing a continuous 50mA of quiescent current. Assume a maximum current draw of 10Apk.
The ripple for a 2m2F/A capacitor bank will be about 0.05A/22mF*~8ms* =~18mVpp between charging pulses. The amplifier PSRR has to attenuate that supply rail ripple sufficiently so that hum is not heard, nor measured at the output of the quiescent amplifier.
When a transient comes along that draws that 10Apk, the ripple will increase massively, to ~3.6Vpp. This is where Current capability assessment comes into the design method.
The amplifier must be able to meet the current demand of the load without the output voltage collapsing in the presence of that 3.6V drop in supply voltage.
* The 8ms between charging pulses allows for a 2ms duration charging pulse giving the 10ms repetition rate for a 50Hz supply. 60Hz supplies will have a proportionately shorter discharge time.
Last edited:
Thanks Andrew,
I didn't understood with my poor knowledge the answer of DF96 fellow (appart it was a "No you can't". My simple understanding with caps is most of time they are able to deliver enough current before voltage beginns to drop ! As a sort of advance of phase ( if I can talk live that, just no other words found) of the current on the voltage )!
Better to have a Pi before all shunts regs, no ? The sense of current in the smoothing caps is always changing very fastly (faster than the voltage In or Out in a reservoir cap I mean).
I beginn slowly ( regarding others design and try to understand keepin apart the too complex things) to "start" my simple shunt stabilizer without measurement ( I imagin here than LTSpice could help me !)
Here is the blocking point : if a low impedance is better to supplu an analog output buffer : a TL431 like or Sipex ref will be better than less noisy diodes which have a greater impedance !
I mean it's better to stay with low impedance while filtering the ref with 2 000 uF low ESR.... even if in the low Frenquencies (< 100 hz ?) even a filtering or a RC filter is not ideal ?
Don't know if I have to follow the work of Christen fellow (link I gave above) about the references : A 15 v zener should be ideal for noise (no problem if I loose 1.2 from the discrete darlington) as the JG buffer worls fine between 9V to 15 V (said to be better near 15 V... better mix Voltage/Current !). If I use a TL431 ON semiconductor instead this will be not for the precision but more for the low impedance !
Think I have my answer for the 6.2 V ref : 4 x red diodes is near with a lower noise. still filtering with a cap it stays quieter than a 6.2 V zener (which has instead the best thermic profile quietness of all zener ?)
I will go with a CSS to feed the voltage ref. Let say a J113 as I don't know otther references !
In the same spirit than my bad question about cap dimension after the bridge for keeping a good dynamic of the load. And if the voltage drop is involved and we still need to have a rapid sourcing of the current to for the load to sound better :
What about a shunt feeding a 10k to 20 k uF cap through a low resistibe like a 0R47/2W to feed the load : cap will see a stable voltage till the 10 K uf cap is enough for fhz above say 100 hz ?
Is it always good bass against good medium-treble when an engineer (not me of course) is designing an analog PS ?
sorry if confusion about concepts...
Charlie (I am today...)
I didn't understood with my poor knowledge the answer of DF96 fellow (appart it was a "No you can't". My simple understanding with caps is most of time they are able to deliver enough current before voltage beginns to drop ! As a sort of advance of phase ( if I can talk live that, just no other words found) of the current on the voltage )!
Better to have a Pi before all shunts regs, no ? The sense of current in the smoothing caps is always changing very fastly (faster than the voltage In or Out in a reservoir cap I mean).
I beginn slowly ( regarding others design and try to understand keepin apart the too complex things) to "start" my simple shunt stabilizer without measurement ( I imagin here than LTSpice could help me !)
Here is the blocking point : if a low impedance is better to supplu an analog output buffer : a TL431 like or Sipex ref will be better than less noisy diodes which have a greater impedance !
I mean it's better to stay with low impedance while filtering the ref with 2 000 uF low ESR.... even if in the low Frenquencies (< 100 hz ?) even a filtering or a RC filter is not ideal ?
Don't know if I have to follow the work of Christen fellow (link I gave above) about the references : A 15 v zener should be ideal for noise (no problem if I loose 1.2 from the discrete darlington) as the JG buffer worls fine between 9V to 15 V (said to be better near 15 V... better mix Voltage/Current !). If I use a TL431 ON semiconductor instead this will be not for the precision but more for the low impedance !
Think I have my answer for the 6.2 V ref : 4 x red diodes is near with a lower noise. still filtering with a cap it stays quieter than a 6.2 V zener (which has instead the best thermic profile quietness of all zener ?)
I will go with a CSS to feed the voltage ref. Let say a J113 as I don't know otther references !
In the same spirit than my bad question about cap dimension after the bridge for keeping a good dynamic of the load. And if the voltage drop is involved and we still need to have a rapid sourcing of the current to for the load to sound better :
What about a shunt feeding a 10k to 20 k uF cap through a low resistibe like a 0R47/2W to feed the load : cap will see a stable voltage till the 10 K uf cap is enough for fhz above say 100 hz ?
Is it always good bass against good medium-treble when an engineer (not me of course) is designing an analog PS ?
sorry if confusion about concepts...
Charlie (I am today...)
Ah, we can understand why a picture is so universal !
Sorry my english ! I try to rephrase my questions.
- For a voltage reference and not speaking about the PSRR, which trade offs between low noise and low impedance when it comes to choose between a TL431 and a zener for an anlog stage need?
- Is a RC filter better than a simple big cap to filtering a voltage reference ?
- Is it not better to feed a huge cap near the load across a resistive low R with a shunt which can provide enough current and enough stable voltage to this huge 10 000 uF cap ?
- In this last case, what are the frequencie where the load will be fed by this last huge cap or the shunt ? I imagine below less 100 hz it will be the emitter follower of the shunt which can fed the load (and not the huge cap because not enough huge and a Z curve which is too high in this range : even with low esr caps, Z is climbing at low and high frqeuncies!)
- I believe to understand there is a sort of tradeoff in each PS design between the quality of the sound between bass and the higher frequencies : e.g. : a stabilize PS is always better in the mediums & highs while a just huge non stabilized PS (just a Pi) is always better with bass in an amp, pre, analog stage (and this is of course a question as believing is not knowing nore understanding)
Sorry my english ! I try to rephrase my questions.
- For a voltage reference and not speaking about the PSRR, which trade offs between low noise and low impedance when it comes to choose between a TL431 and a zener for an anlog stage need?
- Is a RC filter better than a simple big cap to filtering a voltage reference ?
- Is it not better to feed a huge cap near the load across a resistive low R with a shunt which can provide enough current and enough stable voltage to this huge 10 000 uF cap ?
- In this last case, what are the frequencie where the load will be fed by this last huge cap or the shunt ? I imagine below less 100 hz it will be the emitter follower of the shunt which can fed the load (and not the huge cap because not enough huge and a Z curve which is too high in this range : even with low esr caps, Z is climbing at low and high frqeuncies!)
- I believe to understand there is a sort of tradeoff in each PS design between the quality of the sound between bass and the higher frequencies : e.g. : a stabilize PS is always better in the mediums & highs while a just huge non stabilized PS (just a Pi) is always better with bass in an amp, pre, analog stage (and this is of course a question as believing is not knowing nore understanding)
Last edited:
Thanks, DF96.
Ah I was believing (many believing from me when it becomes technical) it was more in DC stable feeding and loading situation but less for the rapid pulse curent demands from the load. In other words I believed the caps was fulled in faster than the load needs in any situation ... and that the drop of voltage around the cap leads was just the consequence of the feeded load already ! (hummm sorry for the English, hope you understand what I try to tell)
So it becomes too much complicate for me if a reference voltage is not enough for sizing the caps after !
And it's not going in the sense of the OP anymore. So I stop here and have to learn more
Ah I was believing (many believing from me
when it becomes technical) it was more in DC stable feeding and loading situation but less for the rapid pulse curent demands from the load. In other words I believed the caps was fulled in faster than the load needs in any situation ... and that the drop of voltage around the cap leads was just the consequence of the feeded load already ! (hummm sorry for the English, hope you understand what I try to tell)So it becomes too much complicate for me if a reference voltage is not enough for sizing the caps after !
And it's not going in the sense of the OP anymore. So I stop here and have to learn more
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.