Good point.
This one's a bit weird (to me) in having no cap after the bridge. I wonder how much you could bring the ripple down by adding a true reservoir cap and rehashing the feed resistors.
This one's a bit weird (to me) in having no cap after the bridge. I wonder how much you could bring the ripple down by adding a true reservoir cap and rehashing the feed resistors.
The first cap is the reservoir cap. This cap always sees some resistance; it's just that in this circuit it sees more than usual. I'm not sure why it is done in this way; maybe the designer wanted to limit charging current peaks, but at the expense of a 'soft' PSU?
Possibly. Even adding 470uf after the bridge (and increasing the series resistors to 270 ohm) sees the ripple come down massively.
Thanks everyone. I'm starting understand this better now.
The schematic mentions that the first four resistors can be swapped for chokes. So I'm guessing the intention was for this to be a choke input supply but I'm not sure. I'm also not sure if a choke input or cap input supply would be best for this application.
At the end of the day this is the PSU for a tube based 2-way active crossover. It will also be used as a buffered, passive volume control for the whole system by adding a volume control to the input. So it's kind of a replacement for a preamp but without any gain.
I want the absolute minimum noise as my speakers are fairly sensitive at 91dB. So I guess I need to keep ripple to a minimum. I'm open to changing this to a cap input supply if the general consensus is that that would be better for the application. I'm also open to swapping any of the first four resistors for chokes if that will yield a significant improvement. I'd have to weigh up the cost though.
Before getting into all that though I'd like to get a better understanding of a few things that I'm still not sure on:
What is the purpose of the two 33k resistors labeled R1 and R14 in the original schematic HERE? Are they to discharge the PSU when power is removed or something else? From what you guys are saying it seems like they're not needed, or at least only needed to provide a load in the simulation and even then only if they're swapped to lower values. So I'm a bit unsure about all that.
Do I need to include some resistors to discharge the PSU when power is removed? If so, what value and where should they go?
Should the 12 lots of 1k resistors and 10uF caps that are on the anode and cathode of the tubes be included in the PSU simulation seen as they form part of the PSU?
Many thanks.
The schematic mentions that the first four resistors can be swapped for chokes. So I'm guessing the intention was for this to be a choke input supply but I'm not sure. I'm also not sure if a choke input or cap input supply would be best for this application.
At the end of the day this is the PSU for a tube based 2-way active crossover. It will also be used as a buffered, passive volume control for the whole system by adding a volume control to the input. So it's kind of a replacement for a preamp but without any gain.
I want the absolute minimum noise as my speakers are fairly sensitive at 91dB. So I guess I need to keep ripple to a minimum. I'm open to changing this to a cap input supply if the general consensus is that that would be better for the application. I'm also open to swapping any of the first four resistors for chokes if that will yield a significant improvement. I'd have to weigh up the cost though.
Before getting into all that though I'd like to get a better understanding of a few things that I'm still not sure on:
What is the purpose of the two 33k resistors labeled R1 and R14 in the original schematic HERE? Are they to discharge the PSU when power is removed or something else? From what you guys are saying it seems like they're not needed, or at least only needed to provide a load in the simulation and even then only if they're swapped to lower values. So I'm a bit unsure about all that.
Do I need to include some resistors to discharge the PSU when power is removed? If so, what value and where should they go?
Should the 12 lots of 1k resistors and 10uF caps that are on the anode and cathode of the tubes be included in the PSU simulation seen as they form part of the PSU?
Many thanks.
The 33k's are as you say, just bleed resistors. In the simulation we've dropped them right down in value to simulate the total load of the amp.
The 1k + 10uf will have a small effect but their primary function isn't ripple rejection. Each 1k will have around 10 volts across it under no signal conditions (due to the valve anode current).
You could add those to the sim as a single lumped value but the effect wont be dramatic.
The 1k + 10uf will have a small effect but their primary function isn't ripple rejection. Each 1k will have around 10 volts across it under no signal conditions (due to the valve anode current).
You could add those to the sim as a single lumped value but the effect wont be dramatic.
The schematic mentions that the first four resistors can be swapped for chokes.
Let us use MOOLY's circuit where the components are ID'd. You're talking about R1, R2 and R3, R4. NOTE that Mooly "got the values wrong" compared to your drawing. Yours was R1 = R2 = R3 = R4 = 330 Ω. His was 820 Ω. Probably just a 'whoops' moment.
So I'm guessing the intention was for this to be a choke input supply but I'm not sure.
No, it wasn't necessarily the INTENTION, but a suggestion for substantially lower hum (better performance.) … at a much, much higher cost (inductors can be pricey.)
I'm also not sure if a choke input or cap input supply would be best for this application.
They are different in that a CAP input maximizes the DC output (all other things equal), whereas a CHOKE input supply (with sufficiently large chokes) averages the DC to something a bit lower than the VRMS value of the transformer secondary windings.
At the end of the day this is the PSU for a tube based 2-way active crossover. It will also be used as a buffered, passive volume control for the whole system by adding a volume control to the input. So it's kind of a replacement for a preamp but without any gain.
OK.
I want the absolute minimum noise as my speakers are fairly sensitive at 91dB. So I guess I need to keep ripple to a minimum.
Good. If price isn't the most important consideration, then I'd recommend BOTH cap and choke. CAP input by changing the R1 and R3 values to 22 Ω (po' boy's rush current limiter), then replacing R2 and R4 with 20 henry chokes. I would also lower the capacitance of C 1 and 2 … to 220 μF or even 100 μF. And in return, I'd add another stage of filtering with a 470 μF set of caps, and another 20 H choke. REALLY quiet. Inaudible hum at 100 ma draw…
I'm open to changing this to a cap input supply if the general consensus is that that would be better for the application. I'm also open to swapping any of the first four resistors for chokes if that will yield a significant improvement.
See above…
I'd have to weigh up the cost though.
Well … you're making a commitment anyway: if $100 in parts is going to break the bank, then there's an entirely different way of solving the power supply, which arguably might even be better. But you have to chuck the tubes-are-gods tinfoil hat, and go with a bipolar regulated supply. Same - or better - result: noiselessness, with added benefit of line (wall power) independent output voltage. I'll run-and-hide in the corner tho' if that offends you.
Before getting into all that though I'd like to get a better understanding of a few things that I'm still not sure on:
Go on…
What is the purpose of the two 33k resistors labeled R1 and R14 in the original schematic HERE? Are they to discharge the PSU when power is removed or something else? From what you guys are saying it seems like they're not needed, or at least only needed to provide a load in the simulation and even then only if they're swapped to lower values.
[1] Safety - to ensure capacitor chain entirely discharges within a reasonable amount of time following power-down.
[2] Minimum load - to ensure that overvoltage is not reached do to NO load.
Should the 12 lots of 1k resistors and 10μF caps that are on the anode and cathode of the tubes be included in the PSU simulation seen as they form part of the PSU?
Well, 'yes and no'. If you already have a solid handle on the 120 ma of circuit current draw, and aren't just pulling the number out of some nether crack… then don't worry about them all. If on the other hand you're working with a design that isn't so well characterized (which might be ¹/₁₀ the current draw!) then again … don't worry about it: one of the things about power supplies which is a universal truth is that the lower the current draw, the better the residual ripple reduction of the supply.
That should answer most of it.
Now… if you haven't yet chucked a Plague infected beaver in my general direction due to mentioning “regulators”, then we might have a different conversation, and a different design diagram. And you could use a few nice glowing gas-discharge voltage regulator tubes too. They're so festive. Attention getters. I like the purple ones (argon). Goes well with satin-finish aluminum and piano-black finishes…
LOL
GoatGuy
Let us use MOOLY's circuit where the components are ID'd. You're talking about R1, R2 and R3, R4. NOTE that Mooly "got the values wrong" compared to your drawing. Yours was R1 = R2 = R3 = R4 = 330 Ω. His was 820 Ω. Probably just a 'whoops' moment.
So I'm guessing the intention was for this to be a choke input supply but I'm not sure.
No, it wasn't necessarily the INTENTION, but a suggestion for substantially lower hum (better performance.) … at a much, much higher cost (inductors can be pricey.)
I'm also not sure if a choke input or cap input supply would be best for this application.
They are different in that a CAP input maximizes the DC output (all other things equal), whereas a CHOKE input supply (with sufficiently large chokes) averages the DC to something a bit lower than the VRMS value of the transformer secondary windings.
At the end of the day this is the PSU for a tube based 2-way active crossover. It will also be used as a buffered, passive volume control for the whole system by adding a volume control to the input. So it's kind of a replacement for a preamp but without any gain.
OK.
I want the absolute minimum noise as my speakers are fairly sensitive at 91dB. So I guess I need to keep ripple to a minimum.
Good. If price isn't the most important consideration, then I'd recommend BOTH cap and choke. CAP input by changing the R1 and R3 values to 22 Ω (po' boy's rush current limiter), then replacing R2 and R4 with 20 henry chokes. I would also lower the capacitance of C 1 and 2 … to 220 μF or even 100 μF. And in return, I'd add another stage of filtering with a 470 μF set of caps, and another 20 H choke. REALLY quiet. Inaudible hum at 100 ma draw…
I'm open to changing this to a cap input supply if the general consensus is that that would be better for the application. I'm also open to swapping any of the first four resistors for chokes if that will yield a significant improvement.
See above…
I'd have to weigh up the cost though.
Well … you're making a commitment anyway: if $100 in parts is going to break the bank, then there's an entirely different way of solving the power supply, which arguably might even be better. But you have to chuck the tubes-are-gods tinfoil hat, and go with a bipolar regulated supply. Same - or better - result: noiselessness, with added benefit of line (wall power) independent output voltage. I'll run-and-hide in the corner tho' if that offends you.
Before getting into all that though I'd like to get a better understanding of a few things that I'm still not sure on:
Go on…
What is the purpose of the two 33k resistors labeled R1 and R14 in the original schematic HERE? Are they to discharge the PSU when power is removed or something else? From what you guys are saying it seems like they're not needed, or at least only needed to provide a load in the simulation and even then only if they're swapped to lower values.
[1] Safety - to ensure capacitor chain entirely discharges within a reasonable amount of time following power-down.
[2] Minimum load - to ensure that overvoltage is not reached do to NO load.
Should the 12 lots of 1k resistors and 10μF caps that are on the anode and cathode of the tubes be included in the PSU simulation seen as they form part of the PSU?
Well, 'yes and no'. If you already have a solid handle on the 120 ma of circuit current draw, and aren't just pulling the number out of some nether crack… then don't worry about them all. If on the other hand you're working with a design that isn't so well characterized (which might be ¹/₁₀ the current draw!) then again … don't worry about it: one of the things about power supplies which is a universal truth is that the lower the current draw, the better the residual ripple reduction of the supply.
That should answer most of it.
Now… if you haven't yet chucked a Plague infected beaver in my general direction due to mentioning “regulators”, then we might have a different conversation, and a different design diagram. And you could use a few nice glowing gas-discharge voltage regulator tubes too. They're so festive. Attention getters. I like the purple ones (argon). Goes well with satin-finish aluminum and piano-black finishes…
LOL
GoatGuy
Great, thanks again.
I'm not 100% sure if the current draw is 120ma. It looks like that info was based on the first version of the circuit which used 12AU7 with conventional resistive cathode loads. This version uses ECC88 with CCS cathode loads. Other than that I think the circuits are basically the same. So I guess the first thing to do would be to establish what the current draw is now.
I intend to very gradually buy parts for this project over several months and I want to do it properly. I don't have an endless budget but I don't mind spending extra if it will make a worthwhile improvement. I'm already considering making this a two box build with the PSU separate to keep noise to a minimum. I'll also consider any recommendations on ways to improve the PSU providing it's not going to costs thousands. I already have plans for elevating the heater voltage which is currently AC in the schematic and I'm happy to stick with that in the initial build to keep things simple.
I'm not a tube snob, so if a solid state PSU is the best option then fair enough. I'm also open to tube regulators providing the tubes are easy to source and relatively inexpensive. The only things I've got reservations over are switched mode PSUs, opamps and electrolytic caps in the signal path as I've found them all to be detrimental. That's why I decided to give this crossover a try as it doesn't use opamps and it gives me the flexibility to tweak the design to add volume control etc so I can ditch my passive preamp and simplify my system.
So what are your recommendations for the PSU?
Thanks.
I'm not 100% sure if the current draw is 120ma. It looks like that info was based on the first version of the circuit which used 12AU7 with conventional resistive cathode loads. This version uses ECC88 with CCS cathode loads. Other than that I think the circuits are basically the same. So I guess the first thing to do would be to establish what the current draw is now.
I intend to very gradually buy parts for this project over several months and I want to do it properly. I don't have an endless budget but I don't mind spending extra if it will make a worthwhile improvement. I'm already considering making this a two box build with the PSU separate to keep noise to a minimum. I'll also consider any recommendations on ways to improve the PSU providing it's not going to costs thousands. I already have plans for elevating the heater voltage which is currently AC in the schematic and I'm happy to stick with that in the initial build to keep things simple.
I'm not a tube snob, so if a solid state PSU is the best option then fair enough. I'm also open to tube regulators providing the tubes are easy to source and relatively inexpensive. The only things I've got reservations over are switched mode PSUs, opamps and electrolytic caps in the signal path as I've found them all to be detrimental. That's why I decided to give this crossover a try as it doesn't use opamps and it gives me the flexibility to tweak the design to add volume control etc so I can ditch my passive preamp and simplify my system.
So what are your recommendations for the PSU?
Thanks.
120ma sounds in the right ballpark for either version. The one with the current sinks seems to have a preset for each stage to allow the current to be set to some predetermined value, although it doesn't seem to say on the circuit what that is.
Something totally unrelated (and I haven't studied the signal circuitry) is that the LF output (and HF) both use 1uf coupling caps feeding into 10k. That gives a 16Hz -3db point which seems very high for something described as an LF output.
I don't normally delve into the design of valve gear but a solid state reg would be the obvious choice to reduce ripple to very low levels. The split supply of this design means you need two regs, a positive and negative one.
Okay... so what ya going to listen to that hasn't been through dozens if not hundreds of both 😀
Something totally unrelated (and I haven't studied the signal circuitry) is that the LF output (and HF) both use 1uf coupling caps feeding into 10k. That gives a 16Hz -3db point which seems very high for something described as an LF output.
I don't normally delve into the design of valve gear but a solid state reg would be the obvious choice to reduce ripple to very low levels. The split supply of this design means you need two regs, a positive and negative one.
Okay... so what ya going to listen to that hasn't been through dozens if not hundreds of both 😀
Read Pete's explanation of the resistors towards the bottom of this page. Why are you trying to simulate a well proven and very simple design?!😕
Thanks.
I emailed Pete (the designer) a few days ago about setting the cathode current and he said to go for 10mA. So that means the total cathode draw will be 60mA. According to the ECC88 datasheet, the anode needs 15mA at 90V. So I guess I can expect a little less current at 100V. Maybe around 13mA. I'd have to do work it out. So I could be looking at more like 78mA for the total anode draw if I've got that right. That would take the total to more like 140mA.
I noticed the bit about bleeder resistors near the bottom of the page which is what made me think that's what the 33k resistors were. I just wanted to check to make sure.
The main reason I wanted to simulate the design was to learn from the experience. I like to learn how things work and how well they work. 🙂
Also, as I mentioned earlier I wanted to experiment with using chokes to see how much difference they made.
I emailed Pete (the designer) a few days ago about setting the cathode current and he said to go for 10mA. So that means the total cathode draw will be 60mA. According to the ECC88 datasheet, the anode needs 15mA at 90V. So I guess I can expect a little less current at 100V. Maybe around 13mA. I'd have to do work it out. So I could be looking at more like 78mA for the total anode draw if I've got that right. That would take the total to more like 140mA.
I noticed the bit about bleeder resistors near the bottom of the page which is what made me think that's what the 33k resistors were. I just wanted to check to make sure.
The main reason I wanted to simulate the design was to learn from the experience. I like to learn how things work and how well they work. 🙂
Also, as I mentioned earlier I wanted to experiment with using chokes to see how much difference they made.
Last edited:
That's how it should be 🙂 Once you understand the basics then it all starts to make more sense and you then have the confidence to start making your own choices.
You can simulate inductors in LT, but I'm afraid I don't know what the typical properties (such as series coil resistance) and even what the typical values would be that you would use for this. What is readily available inductor wise. I'm much more at home with solid state 😀
All these power supplies are unregulated which means that the output can vary both as the load changes and also due to changes in mains voltage. A true regulator would remove those variables and as a bonus significantly reduce the ripple as well.
- Status
- Not open for further replies.