sci.electronics.design | Cheap AndEasy 6.3 Volts| page 4

Reply by ●February 5, 20182018-02-05

>"What it yields depends entirely on the size of the cap. "

Still that is two components instead of one. And if you use a cap like that it is not advisable to use a regular lytic. Something like a speaker crossover cap would be more suited, though in this application does not need to be bipolar.

It works now, all I have to do now is to goop the wall wart to  something inside the cabinet, out of the way hopefully.

Reply by ●February 5, 20182018-02-05

On Monday, 5 February 2018 15:46:08 UTC, jurb...@gmail.com  wrote:
nt:

> >"What it yields depends entirely on the size of the cap. "
> 
> Still that is two components instead of one. And if you use a cap like that it is not advisable to use a regular lytic. Something like a speaker crossover cap would be more suited, though in this application does not need to be bipolar.

why? Do you expect to exceed ripple current spec at 100hz?

> It works now, all I have to do now is to goop the wall wart to  something inside the cabinet, out of the way hopefully.

yep!


NT

Reply by ●February 5, 20182018-02-05

>"Do you expect to exceed ripple current spec at 100hz? "

Filters are supposed to filter. They are not supposed to be used as reactive components. First of all the tolerance is too loose, the capacitance can be double he rating and it is a pass. So you really can't predict what you're going to get.

Crossover caps have an AC rating. I am not sure about the ripple current issue but it seems the value is low enough that it might be an issue. But I do know you can't use regular lytics for a crossover. They won't last.

And if you think about it, why not just use such an AC cap in
series and let its Xc do the job ? With rectification it would be about the same anyway. But one resistor suits me just fine.

Actually if it was a high frequency switched mode I could capacitor couple the output to the filament and use caps that are rated more than 3,500 volts DC. Then the transformer doesn't have to take any DC.

But this is all academic. What was a bit surprising is the filament current. It must be around 600 mA which I think is high. Perhaps they have a different coating that needs to be hotter to work. If not, it seems like it wouldn't last all that long.

But I got more fish to fry. I got one channel clipping and the other deflected clean off the screen. Next is a quick look at the power supply, see if it is regulating. I learned the hard way that in a scope when they say 15.5 volts they mean 15.5 volts, not 17, not 14, 15.5. Somehow, even though the circuit is completely complementary a voltage that is off can cause an offset. Maybe because of slight differences in the transistors. I should find more suitable ones. The originals have a gain of 5 and a bandwidth of 300 MHz, the ones I stuck in have a gain more like 200 but the bandwidth is low, something like 2 MHz. Its intended application was audio. Hell, they use a higher bandwidth transistor for audio now. In fact I do't know if you can even get one with a bandwidth that low.

Of more concern would be finding one with gain that low. And I don't know of any way to decrease effective hfe in a transistor. Any bright ideas on that ?

i wonder what is the lowest current gain transistor available now. i know some of the switching transistors are that low but those are not linear so they are probably not suitable. You should have seen when I tried to use a switcher as a video amp. The gamma was off in a funny way.

Luckily it is used common collector (don't say it) so high hfe probably is not a concern, but not THAT much. 200 vs 5 ? I don't hold much hope for it working right. I was just looking for DC balance, getting rid of the clipping. That would tell me nothing alse is wrong. But I consider the indications to be inconclusive.

Reply by Phil Allison ●February 5, 20182018-02-05

jurb...@gmail.com wrote:

-------------------------------
>
> >"Do you expect to exceed ripple current spec at 100hz? "
> 
> Filters are supposed to filter. 
>

** Power supply "filtering" refers to removing *supply frequency* ripple from a rectified AC wave in order to create a smooth, DC rail. Capacitors do the job of SMOOTHING by storing a small amount of charge when the voltage is at it highest and delivering it back when the voltage drops away.  

The process is fundamentally different to linear filters used in speaker crossovers etc.

In fact, caps used in DC supplies are often called "smoothing caps". 

> Crossover caps have an AC rating.I am not sure about the ripple current 
> issue but it seems the value is low enough that it might be an issue.
> But I do know you can't use regular lytics for a crossover. 

** And crossover ( bi-polar) electros have no place in DC supply smoothing. 
> And if you think about it, why not just use such an AC cap in 
> series and let its Xc do the job ?

** Cos that will not allow you to boost a voltage, only drop it. 

> 
> Actually if it was a high frequency switched mode I could capacitor couple
> the output to the filament and use caps that are rated more than 3,500 
> volts DC. Then the transformer doesn't have to take any DC. 
> 

** Many scopes use DC to DC inverters instead of a full SMPS. 

A standard iron core mains transformer feeds a linear PSU using regulator ICs to produce positive and negative rails of 15 to 20VDC. This DC supply runs a most of the internal circuitry including an inverter, running at about 30kHz, that drives a ferrite transformer with a square wave.

The secondary of the ferrite transformer has multiple windings covering the high voltage needs of the scope. One winding, with only a few turns, supplies 6.3V to CRT. 

Because of the use of square waves, DC rails derived from these windings need very little filtering or smoothing, plus they are all *regulated* since the DC supply powering the inverter is regulated. 

The scheme is simple, reliable and rivals or betters SMPS efficiency. 

My own 50MHz dual trace scope is built like this, its total power consumption is 14.5 watts. 

....   Phil