Hello aR,
I've had excellent results using passive filtering before my regulator. CLCLC filtering, knocking down the 21Vdc LV rail to around 11Vdc. This also allows substantially less capacitance in the rail and lower ripple....as we know from building HV rails for tube gear.
The LV chokes are custom wound for me by Electra-Print.
I too plan to get the Hypnotize board from Tea-Bag.
W
Hi there, Wheeze!...
Hi there Salas,
CLC filtering is great - although, as you like to say, CRC has the disadvantage of being the poor man's CLC, but it also has the... well, advantage of being the poor man's CLC, if you know what I mean... Saved funds can always be put to good use. Will get to using L's when everything is nailed down in detail in the DDDAC's PS. As I do agree that chokes are a must in filtering.
Anyway - and this is for Salas - concerning calculations; I thought you said in your writeup that the CCS has to provide about twice the actual current draw of the load. Having planned to use (16) 1543 chips, my current draw (actual) would be around 800-850mA. Therefore the CCS should be able to provide (and adjusted to) 1.6-1.7Amps - is that correct?
This current then gets split between the load and the shunt MOSFET - right? - and for instance, in case the load gets detached, the shunt MOSFET, in case the regulator is powered at that time, has to be able to handle all of the CCS MOSFET heat load, as it handles the full CCS current, right?
This is the reason I will provide identical heatsinks to both MOSFETs.
Secondly, this is also the reason the CCS needs to be adjusted to provide, say, 1.8Amps - correct?
Thanks much,
Radu.
Hi there Salas,
CLC filtering is great - although, as you like to say, CRC has the disadvantage of being the poor man's CLC, but it also has the... well, advantage of being the poor man's CLC, if you know what I mean... Saved funds can always be put to good use. Will get to using L's when everything is nailed down in detail in the DDDAC's PS. As I do agree that chokes are a must in filtering.
Anyway - and this is for Salas - concerning calculations; I thought you said in your writeup that the CCS has to provide about twice the actual current draw of the load. Having planned to use (16) 1543 chips, my current draw (actual) would be around 800-850mA. Therefore the CCS should be able to provide (and adjusted to) 1.6-1.7Amps - is that correct?
This current then gets split between the load and the shunt MOSFET - right? - and for instance, in case the load gets detached, the shunt MOSFET, in case the regulator is powered at that time, has to be able to handle all of the CCS MOSFET heat load, as it handles the full CCS current, right?
This is the reason I will provide identical heatsinks to both MOSFETs.
Secondly, this is also the reason the CCS needs to be adjusted to provide, say, 1.8Amps - correct?
Thanks much,
Radu.
Rax:
In small currents its good to double. As long as you will have 200 to 300mA spare on your shunt it will be in a very good point. Don't go doubling when in A territory. The heat creeps up hard. I was talking 2R7 when you will be running 8 chips for start. Plan for 1.2A when with 16 chips. The spare will be ample and the heat logical. I guess its better not to plan sinking for when running it with no load, increases you bulk and cost. Its cheaper to just turn it off.
In small currents its good to double. As long as you will have 200 to 300mA spare on your shunt it will be in a very good point. Don't go doubling when in A territory. The heat creeps up hard. I was talking 2R7 when you will be running 8 chips for start. Plan for 1.2A when with 16 chips. The spare will be ample and the heat logical. I guess its better not to plan sinking for when running it with no load, increases you bulk and cost. Its cheaper to just turn it off.
PCB for half/hypnotize
Guys,
Been thinking about something, although I don't want to promise anything yet, but here it is. I've been thinking on laying out a PCB for what I need this to be (single rail higher amperage PS, timer delay included), and was wondering if you guys (Salas) would mind, and if you'd agree to help me designing this, and possibly giving me feedback when I can show something done and ready for review.
Would that be OK, as far as you're concerned? If I get my act together, I'd probably CAD something up this weekend. Just that you know, I'm very rusty, as last time a executed a PCB design/build, I was 16 and was doing this very very low tech. It's true, it was my N-th PCB, but still.
I just want to make sure that first, I don't disrespect some ownership aspects, and second, that you would also maybe agree to give feedback.
Please let me know - thanks.
Best,
Radu.
Guys,
Been thinking about something, although I don't want to promise anything yet, but here it is. I've been thinking on laying out a PCB for what I need this to be (single rail higher amperage PS, timer delay included), and was wondering if you guys (Salas) would mind, and if you'd agree to help me designing this, and possibly giving me feedback when I can show something done and ready for review.
Would that be OK, as far as you're concerned? If I get my act together, I'd probably CAD something up this weekend. Just that you know, I'm very rusty, as last time a executed a PCB design/build, I was 16 and was doing this very very low tech. It's true, it was my N-th PCB, but still.
I just want to make sure that first, I don't disrespect some ownership aspects, and second, that you would also maybe agree to give feedback.
Please let me know - thanks.
Best,
Radu.
I only make p2p so I am bad on designing boards.
But its a simple CCT and I think you can make it.
There is some great work shared early in the shunt thread by Jameshillj you may copy though. James's PCB work.
Silk screen side only
But its a simple CCT and I think you can make it.
There is some great work shared early in the shunt thread by Jameshillj you may copy though. James's PCB work.
Silk screen side only
No, it will not help. Can you find a 100uF BP (bipolar) lytic like Nichicon Muse or cheaper for output, and just 0.1uf to 1uF axial MKT or some axial MKP which is small enough to fit there upright across Vref in the place of the lytic? That would up your resolution maybe. Its a trade off between some noise filtering and cap distortion. Leds are low noise though and this is line level. Their impedance is low, the VLF won't be affected. You tell us what you can hear.
PCBs
Hello everybody,
After one exhausting day, here's my (first, admittedly rough around the edges) take on a single rail hypnotize with timer delay... Please comment away, don't hold back.
The rather baroque filtering scheme is due to some extensive listening sessions from Wheezer (hello there) on the DDDAC1543MKII that we intend to use this for. I can fairly easily adjust the filtering section per simpler approaches.
I am still not sure about the grounding scheme, I initially kept all higher current traces at 1.5mm width, including all ground, but then I enlarged all ground traces freely wherever I could do that easily.
Thank you for your input in advance.
Thank you,
Radu.
Hello everybody,
After one exhausting day, here's my (first, admittedly rough around the edges) take on a single rail hypnotize with timer delay... Please comment away, don't hold back.
The rather baroque filtering scheme is due to some extensive listening sessions from Wheezer (hello there) on the DDDAC1543MKII that we intend to use this for. I can fairly easily adjust the filtering section per simpler approaches.
I am still not sure about the grounding scheme, I initially kept all higher current traces at 1.5mm width, including all ground, but then I enlarged all ground traces freely wherever I could do that easily.
Thank you for your input in advance.
Thank you,
Radu.
Howdy,
Not a big fan of Nichicon - I have a homebrew ESR meter (@100kHz) that I went around my parts department once with, and decided the clear winners in the low-ESR category: third place - Elna Cearfine; second (on tie with first) Panasonic FM, and first: Black Gate N series... Nichicon didn't make it near the podium with any model, even KZs trailed the Pannies FM by a pretty good shot.
In any case, I favor the Pannies FM for an exceptional price/availability/performance ratio. But they are only available in lower values (for instance, in 25V flavor, they only go as high as 2700uF, and they are not ready to order from Digikey), so I decided to provide the possibility of putting more of them if deemed necessary.
I didn't get the chance to ask more questions about the timer delay, but I thought it's been provided in all of the finished/released PCBs, including hypnotize and mesmerize. Isn't it? Also, it's one strange timer delay, can you tell me more about its workings? I get the basics (the junction opening when cap charges to value etc.), but whassup with the other transistor used as diode, I guess, as only one junction is placed in circuit?
The chokes are intended to be the H's range and are likely/expected heavy (don't necessarily want three, though...). They're intended to suppress ripple towards the hundreds of uV range. Right now I get ripple (with a Sjostrom/Jung regulator and decent filtering) in the 11-13mV range. I want that towards 100uV or less. I assume the CCS has a high PSRR, so the chokes might be needed or not - will see.
The heatsinks are the best I could source from Aavid Thermalloy that can be of use to TO247 Ts, but they're around 2.7C/W... Hoipefully they'll do the trick.
Thanks for your feedback!
- Radu.
Not a big fan of Nichicon - I have a homebrew ESR meter (@100kHz) that I went around my parts department once with, and decided the clear winners in the low-ESR category: third place - Elna Cearfine; second (on tie with first) Panasonic FM, and first: Black Gate N series... Nichicon didn't make it near the podium with any model, even KZs trailed the Pannies FM by a pretty good shot.
In any case, I favor the Pannies FM for an exceptional price/availability/performance ratio. But they are only available in lower values (for instance, in 25V flavor, they only go as high as 2700uF, and they are not ready to order from Digikey), so I decided to provide the possibility of putting more of them if deemed necessary.
I didn't get the chance to ask more questions about the timer delay, but I thought it's been provided in all of the finished/released PCBs, including hypnotize and mesmerize. Isn't it? Also, it's one strange timer delay, can you tell me more about its workings? I get the basics (the junction opening when cap charges to value etc.), but whassup with the other transistor used as diode, I guess, as only one junction is placed in circuit?
The chokes are intended to be the H's range and are likely/expected heavy (don't necessarily want three, though...). They're intended to suppress ripple towards the hundreds of uV range. Right now I get ripple (with a Sjostrom/Jung regulator and decent filtering) in the 11-13mV range. I want that towards 100uV or less. I assume the CCS has a high PSRR, so the chokes might be needed or not - will see.
The heatsinks are the best I could source from Aavid Thermalloy that can be of use to TO247 Ts, but they're around 2.7C/W... Hoipefully they'll do the trick.
Thanks for your feedback!
- Radu.
Bipolars are better, and can come from any brand you favorite. For best cap distortion that is, use NP caps across the Vref and output.
The delay is for the audio part's output, not for the rail, because its a DC coupled circuit that bumps a bit during switch off mainly. Skip it. The Hypno as a reg only does not utilize it either.
Large chokes in the H territory are going to have enough capacitance across turns so to render their very high frequency capabilities only theoretical, plus hundreds Ohm. Mini chokes can filter where the CCS stops still, they are elegant, cheap and will not add much resistance to slow down your low ESR caps. Else, low ESR goes off the window.
Remote sensing I highly recommend since you make your own board. Arrange and wire as in the example I posted for you. 4 wires meet at the load. If not closed at their ends together, reg will not work.
The delay is for the audio part's output, not for the rail, because its a DC coupled circuit that bumps a bit during switch off mainly. Skip it. The Hypno as a reg only does not utilize it either.
Large chokes in the H territory are going to have enough capacitance across turns so to render their very high frequency capabilities only theoretical, plus hundreds Ohm. Mini chokes can filter where the CCS stops still, they are elegant, cheap and will not add much resistance to slow down your low ESR caps. Else, low ESR goes off the window.
Remote sensing I highly recommend since you make your own board. Arrange and wire as in the example I posted for you. 4 wires meet at the load. If not closed at their ends together, reg will not work.
It's used in reverse mode.
Veb0 is specified as 5V max. This tells us the diode will hold 5V and not go into avalanche mode.
It does not tell us when it will go into avalanche mode. It's effectively a Zener Diode of unspecified value other than we are told it's >5V.
A properly specified Zener would be better.
I used a bc550b in conventional transistor mode to effectively create a darlington with the second bc550b. It needs an extra resistor to feed the collector. Rather than use an integrated Darlington.
I also run 15V reg through 2K to feed the 1k 12V relay. And it switches without fail. Yes, the 12V relay has only 5V on it.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.