Alright ill be digging into spice then. I have my habits after all I spend a lot of time in "eagle" with my controll prefferences. I hope it should be no problem setting it up once I find out what does what.
That team idea sounds so reallistic
.I will have to somehow figure out how other topologies work. All I know how to handle is flyback and reallistically we dont want a flyback power supply thats over 100W.
As in terms of switching frequency, I get it. Going high on frequency is good because power density is going up aswell but you have to consider mains regulations. That means stuff above 50kHz is regulate by guidelines. I am sure you know this but I had to mention it. Its essentially the reason why I never did want to go above 50kHz
I have done my homework quite a while ago. And thats where I got my current transformer from (LT3751). However easy the circuit looks on paper i tried to build it and bought the 20euro chip aaaaand it didnt work. The inconvenience of a horrible to solder package was also a part contribution to the failure.
I fear this needs a complete ground up design and once we figure out with a discrete circuit what we actually want then I can go arround looking for a integrated solution that will fit my wants. Regadless thanks for the contribution
if youre proposing to rip out the entire power supply and reverse engineer it I refuse that. Not only is it killing the purpose of this project but its a insane waste of time trying to modify a circuit to our wants than creating one to fit our wants. I learned that the hard way..
Last edited:
There is yet another very simple and intuitive way of achieving a high tension DC power supply, and this is by using a sine generator at a frequency much higher than the mains frequency to reduce the transformer size.
A frequency like 20kHz is 20000/50 = 400 times bigger than the mains frequency as used in Europe. In other parts of the world is can be lower, but still, very high. A frequency like this reduces the required self inductance of a primary coil of an HT transformer by 400 for the same magnetising current! The number of turns in the primary is reduced by sqrt(400) = 20 times. With a few turns as a primary and a coil of many more turns as the secondary this should be more straight forward to achieve compared to a high voltage SMPS.
Building a 50kV DC power supply.
A frequency like 20kHz is 20000/50 = 400 times bigger than the mains frequency as used in Europe. In other parts of the world is can be lower, but still, very high. A frequency like this reduces the required self inductance of a primary coil of an HT transformer by 400 for the same magnetising current! The number of turns in the primary is reduced by sqrt(400) = 20 times. With a few turns as a primary and a coil of many more turns as the secondary this should be more straight forward to achieve compared to a high voltage SMPS.
Building a 50kV DC power supply.
yet again... its basically what a smps does - overclock the AC voltage (alternating DC in case of a SMPS) for only having a keybonus in smps is that there is voltage regulation. Also I dont see a easy way of creating a high power high voltage sine wave. T to make thing difficult for efficiency sake it would have to be a class D amplifier feeding a transformer and above that for a 20khz sine wave we would need a 2X of that frequency to replicate it properly in the D amplifier (smaple frequency) and you also need to generate the sine 20khz rather acurately. Does it sound simple? definitely not simpler than building a smps. Not to insult you but its like scratching your left ear with your right hand..possible but why.
I do not see a LVD issue. The LVD doesn't apply to internal equipment voltage. No one will leave the board laying on their sofa, it will be in an amp. Also, the proposed LV supply would be in the same box, and is under 75Vdc and has it's own CE certification.
My 10V rail is for heaters, and if the regulation on the line is better than 10% then the voltage spec can be relaxed to 6.3V and the current demand can be lessened to 9A to put less strain on the transformer design.
Getting a legal CE label requires sending the unit to a testing lab and probably a lawyer. I see big bucks for both and long delays and paperwork.
In my mind I see two plans for this. It is a DIY solution for a hobbyist, or an OEM puts it in another device for retail sale. Neither would require any additional certifications.
The OEM's new device might, when taken as a whole, but not the board itself.
My 10V rail is for heaters, and if the regulation on the line is better than 10% then the voltage spec can be relaxed to 6.3V and the current demand can be lessened to 9A to put less strain on the transformer design.
Getting a legal CE label requires sending the unit to a testing lab and probably a lawyer. I see big bucks for both and long delays and paperwork.
In my mind I see two plans for this. It is a DIY solution for a hobbyist, or an OEM puts it in another device for retail sale. Neither would require any additional certifications.
The OEM's new device might, when taken as a whole, but not the board itself.
Hi,
I admit I had difficulties to get our design going at first, but that was due to a colleague´s design errors which required extensive redesign.
But if one sticks to the datasheet -which indeed needs to be studied carefully- and the recommended parts the LT3751 works really well ... and it costs about 8-9€/1pcs at mouser.
But anyway, proceed as you wish.
jauu
Calvin
I admit I had difficulties to get our design going at first, but that was due to a colleague´s design errors which required extensive redesign.
But if one sticks to the datasheet -which indeed needs to be studied carefully- and the recommended parts the LT3751 works really well ... and it costs about 8-9€/1pcs at mouser.
But anyway, proceed as you wish.
jauu
Calvin
Mouser prices are truly decieving. The delivery service is 50 euros. Not so cheap anymore.
Voltage regulation of the output from an SMPS is a direct consequence of having a negative feedback loop controlling power feed into the transformer.
With all said, I still think if you can do the powersupply properly from mains the stick to it.
Regarding your heater voltage:
I see it this way- There will be two separate circuits one to do the HV line which will have lots of dynamic load on and ther one will be takinc care of the 6,3V (orwhatever V) and the bias voltage at least -100V. Since heaters are damn near a constant load we wont have fluctuating bias voltages and above all we want the bias voltage to come up first and stay on the tubes even if no HV is applied. SO even if the tube is sitting the cold just heating up the cathode before service it already has negative grid voltage ready to go
Id say its more difficult to not only make a perfect sinewave stable but to modulate it in aswell and the controll the output voltage witha NFb by somehow putting another modulation thing inbetween? The class D amplifier make it already too complicated let alone feedback
CE Marking: CE mark approval, CE testing, CE mark requirements, how achieve them all for my product?
Do you really want to go down this rabbit hole? Your choice. If this was an item with broad sales appeal, I'd say go for it.
Do you really want to go down this rabbit hole? Your choice. If this was an item with broad sales appeal, I'd say go for it.
remember you still need isolation. Most of the time preamps are the main point of ground in audio circuits. If youre using a regular ATX thats ground is mains ground refferenced youll get all sorts of bells and whistles from your amplifier (and thats not in a good way)
What transformer? In that cas youre expecting it being isolated well its not as simple as hooking negative feedback on. You need a TL431 and optocoupler setup.
In a push-pull boost SMPS, there is a step up transformer, driven on the primary by a pair of MOSFETS, controlled by the control chip. In the example I described, an SG3525. You can use an optocoupler, but I think I've seen just a voltage divider with the output going to the error amp input. The SG' has a built in 5.1V reference. I'm not saying this is the best control chip, but, it is cheap, readily available and seems easy to use. I admit I may be missing something in its use, but the issues I've seen so far, relate to using it in either half or full bridge. Some tweaking is probably needed to use it at 100k, but it is designed to work up to 300k.
no you dont quite get what I mean. For you to be able to sense the a voltage with that ic it has to be on the same ground potential. For driving the fets you also need something on the same ground potential.
There are two ways o tackle this problem. Offsite regulation: basically the IC is on the secondary side and is driving the primary fets trough a fet drive transformer which does require a sophisticated startup or primary regulation which is much easier I think, with a optocoupler to provide feedback.
There are two ways o tackle this problem. Offsite regulation: basically the IC is on the secondary side and is driving the primary fets trough a fet drive transformer which does require a sophisticated startup or primary regulation which is much easier I think, with a optocoupler to provide feedback.
as you can tell thats not galvanically isolated...but this becomes only a problem is youre using a ATX powersupply or any power supply that has a grounded case.