9 V To 48 V DC-DC Converter Design

[Samuel Groner]

I did notice that you might want to add some input bypass, possibly a C, C/RFC, or a full pi C/RFC/C network at the input.

Can you elaborate why you think I'd need an input filter? This thing is powered from a closely located battery which's not used for anything else.

Also might want to add a bypass capacitor at the ref input to the TLV3401.

The "ref pin" is the inverting or noninverting input?

Would it be worthwhile to add an LED on the 48 V side as well? I know that you probably want to minimize current drain but it would be, in operation, a good confidence-booster to see.

I'll think over it.

Samuel

 

[jdbakker]

[quote author="Samuel Groner"]I drew a more detailed schematic of the current plan: 48V_box_r1.pdf

May I ask for a quick sanity check before I start breadboarding?[/quote]
For some reason I didn't get a Topic Reply Notification for this one, so sorry for the slow reply.

The converter itself looks OK. A capacitor on the input isn't strictly necessary, although if it were my design I'd add a 100n ceramic with possibly a 100u 16V in parallel to dampen resonations on the battery wire (since otherwise it'll start acting as part of the switcher's inductor).

I don't see what the TLV3401 is meant to achieve. The '3401 is an open-drain part, so the LED will never light as drawn. Should this be a low-battery detector ?

Finally, two mikes attached to P48 draw 28mA max. At 100% conversion efficiency, that's 150mA @9Vin, so that fuse may be a bit underrated.

JDB.

 

[Samuel Groner]

Thanks for your answers.

I don't see what the TLV3401 is meant to achieve. The '3401 is an open-drain part, so the LED will never light as drawn. Should this be a low-battery detector?

Should be a low-battery detector. So do I need a pull-up resistor to make it light? Or am I missing something more serious?

Finally, two mikes attached to P48 draw 28 mA max. At 100% conversion efficiency, that's 150 mA @ 9 Vin, so that fuse may be a bit underrated.

Oh sure, I thougth the wrong way around (lower voltage = lower current :roll: )... The phantom power standard sets maximum current draw to 10 mA/mic though, IIRC.

Samuel

 

[Samuel Groner]

Bump--anyone who can enlighten me on the proper use of the given comparator here?

How 'bout putting a rev polarity shunt diode to blow the fuse.

:thumb:

Samuel

 

[mcs]

And a small add-on question, if nobody minds...

What would I have to change in Samuel's circuit to power up to 10 mics?

I will have +/-15V available from the existing power supply.

Best regards,

Mikkel C. Simonsen

 

[Samuel Groner]

I have a variation on my design in my mind which would fulfil your needs, I think. I can draw it if you like.

Samuel

 

[mcs]

[quote author="Samuel Groner"]I have a variation on my design in my mind which would fulfil your needs, I think. I can draw it if you like.[/quote]
Yes, plaese do! I have dabbled in many things over the years, but switchmode supplies is not one of them...

Best regards,

Mikkel C. Simonsen

 

[jdbakker]

[quote author="Samuel Groner"]anyone who can enlighten me on the proper use of the given comparator here?[/quote]
Depends on what you want. This comparator can only sink current due to its open drain output, so no matter what else, the LED/resistor needs to go between +9V and the comparator output.

If you want the LED to light when the battery is >6V, keep U102A connected as-is.

If you want the LED to light when the battery is <6V, swap the comparator's inputs (+in to +9V, -in to R107/R108).

Me, I don't like to rely on an almost-empty battery to supply enough current to light a LED. Plan C would be to connect a 10uF capacitor to the positive supply, with a 220k resistor from its other terminal to ground. Connect this RC to the -in of U102A, and keep the +in connected as-is in the schematic. This will 'blip' the LED on turn-on if the battery is >6V, assuming the +48V ramp-up time is faster than 1s. Hook a reverse-biased diode across the 220k resistor to protect the input of U102A from the cap's discharge when the power is turned off.

[quote author="mcs"]What would I have to change in Samuel's circuit to power up to 10 mics?

I will have +/-15V available from the existing power supply.[/quote]
Ten mics is 140mA max, for a ~600mA switch current @15V (with Ipeak even higher, depending on the inductor used). The 2N7000 can't quite handle that; I'm not even sure that it would be wise to use the LT1073 in that scenario. Better to use a switch controller chip with an external hi-voltage FET, like the LT1738.

JDB.

 

[edanderson]

[quote author="Samuel Groner"]Linear recommends the Sanyo OS-CON type, but I did not find a source for them. [/quote]

here in the states, they're available from b&d enterprises (purveyor also of sanyo semiconductors):

http://www.bdent.com

vishay also makes an organic polymer electrolytic line which is available through mouser (page 583). be warned, they are not cheap.

ed

 

[Samuel Groner]

Thanks for your answers! I updated the schematic to incorporate the latest additions.

This comparator can only sink current due to its open drain output, so no matter what else, the LED/resistor needs to go between +9 V and the comparator output.

I see, makes sense.

Me, I don't like to rely on an almost-empty battery to supply enough current to light a LED. Plan C would be to connect a 10 uF capacitor to the positive supply, with a 220k resistor from its other terminal to ground. Connect this RC to the -in of U102A, and keep the +in connected as-is in the schematic. This will 'blip' the LED on turn-on if the battery is >6 V, assuming the +48 V ramp-up time is faster than 1 s. Hook a reverse-biased diode across the 220k resistor to protect the input of U102A from the cap's discharge when the power is turned off.

Cool! :thumb: Did I got it right on the new schematic?

Here in the states, they're available from B&D enterprises (purveyor also of sanyo semiconductors).

For this project ordering a single capacitor from the USA is out of reach financially, so I settled with simply using a standard low impedance part. Good to now for the futere nonetheless.

Samuel

 

[datape]

This is kind of connected. Can anyone recommend a way to get +24VDC from +15VDC?

Could I just use a 317 regulator set to 24V output and a few pump charge capacitors or would this cause the regulator to misbehave?

Or would I need a dedicated chip like the LT1073?

Thanks

 

[BradAvenson]

[quote author="datape"]This is kind of connected. Can anyone recommend a way to get +24VDC from +15VDC?[/quote]

Fast and easy would be something like the DC-DC's on this mouser page:
http://www.mouser.com/catalog/628/1636.pdf

The NMA series has a 15V DC input to +/-12V out. Since the ouput is isolated you could use that as a +24v rail.

 

[Samuel Groner]

Here's my suggestion for the 50 mA version: [removed, see first page]

It uses a linear post regulator to provide a low impedance output and for very low ripple. Could someone have a look at this before Mikkel runs into problems? Especially the choice of Q1 and C4?

BTW, I think I need to change the switching diode D104 in the original plan as well as the 1N5818 is not rated for 48 V.

Samuel

 

[datape]

Cheers. I'll try something similar..


[quote author="BradAvenson"][quote author="datape"]This is kind of connected. Can anyone recommend a way to get +24VDC from +15VDC?[/quote]

Fast and easy would be something like the DC-DC's on this mouser page:
http://www.mouser.com/catalog/628/1636.pdf

The NMA series has a 15V DC input to +/-12V out. Since the ouput is isolated you could use that as a +24v rail.[/quote]

 

[jdbakker]

[quote author="Samuel Groner"]Here's my suggestion for the 100 mA version: 48V_DC-DC_100mA_r1.pdf

It uses a linear post regulator to provide a low impedance output and for very low ripple. Could someone have a look at this before Mikkel runs into problems? Especially the choice of Q1 and C4?[/quote]
Q1 is fine, although a smaller FET (with less gate charge) would be better. That 3mA shunt regulator needs to power the charging/discharging of the gate capacitance too; my brain fails to produce the relevant formulas ATM, sorry.

I would really suggest not using that LT1073. Even at only 100mA output you're pushing the switch peak current and package power dissipation to their Abs Max ratings and beyond. It might work, it might blow up during a critical session with several power-hungry condenser mics. A properly dimensioned switcher will be much more reliable; I'd not use the '1073 past 50mA @48V.

JDB.

 

[Samuel Groner]

Q1 is fine, although a smaller FET (with less gate charge) would be better.

I did look for a smaller part but didn't find anything convincing--do you have a suggestion?

Even at only 100 mA output you're pushing the switch peak current and package power dissipation to their Abs Max ratings and beyond.

If my math doesn't suck I get about 700 mA peak switch current which would be below the maximum rating (didn't check the package dissipation--how do I calculate that?). I agree with you that if this were a 100 mA continuous supply it would be wise to switch to a different solution, but I guess in real world we would hardly face anything more than 40 mA. Depends on the application though, so Mikkel would need to comment on this.

Samuel

 

[jdbakker]

[quote author="Samuel Groner"][quote author="I"]Q1 is fine, although a smaller FET (with less gate charge) would be better.[/quote]
I did look for a smaller part but didn't find anything convincing--do you have a suggestion?[/quote]
The BS170 looks good on paper.

[quote author="Samuel Groner"]If my math doesn't suck I get about 700 mA peak switch current which would be below the maximum rating (didn't check the package dissipation--how do I calculate that?).[/quote]
Package power dissipation is tricky to calculate, since the switch voltage depends on switch current, so you'd need to factor that into the equation. As a rough estimate, the switch current is a sawtooth going from zero to Ipeak, and the switch voltage is 0.8...1V. Thermal runaway is a real danger, since dissipation increases temperature, which increases switch voltage, which increases dissipation.

[quote author="Samuel Groner"]I agree with you that if this were a 100 mA continuous supply it would be wise to switch to a different solution, but I guess in real world we would hardly face anything more than 40 mA. Depends on the application though, so Mikkel would need to comment on this.[/quote]
Like I said, it might work fine, it might blow up. For ten low-current mikes there will likely never be a problem.

Maybe I'm too much of a defensive driver, but I would never design a supply where a viable worst-case load might come close to exceeding a supply's thermal envelope. [The Behringer ADA8000 thread makes a nice case in point]. Sure, you/Mikkel may have only low-current microphones right now, but what happens a few years down the line when you hook up a set of different microphones ?

At the risk of being a nag, why do you want to use the LT1073 ? It was designed with 1/2/3-cell battery step-up applications as a target, not low-noise power from regulated rails. The fact that it only supports discontinuous operation is quite bad for output noise and EMI. It might make sense for two microphones from a 9V battery if the '1073 is all you can lay your hands on, but not for 10 mikes from a hard 15V rail. Sure, if you really want to you can get around the package power limitation (e.g. by using a non-bootstrapped version of the 3V-6V@1A converter on page 14 of the datasheet), but it's not the right tool for the job, not by a long stretch.

The fixed input voltage and linear post-regulator eliminate most of the 'hard' parts of a SMPS. If you want an easily-available IC, I suspect you can get similar performance with a 555.

JD 'again, left as an excersize to the reader' B.

 

[Samuel Groner]

The BS170 looks good on paper.

Thanks. Aren't there any TO-225 MOSFETs, BTW? Initially I looked for such a part.

Maybe I'm too much of a defensive driver, but I would never design a supply where a viable worst-case load might come close to exceeding a supply's thermal envelope.

Perhaps you're right (I surely wouldn't sell such a thing), so I'll rescale the note on the schematic for the current draw to 50 mA. I'm too busy to dig through the pages to design something with a LT1738 at the moment, so I cannot offer more as a solution to Mikkels design task right now.

Samuel

 

[Samuel Groner]

I updated both files linked in the first post. Both include some detail changes.

Samuel

 

[Samuel Groner]

Finally we got around breadboarding this thing (i.e. the 20 mA version). It works well, noise is relatively high as expect (40 mVpp unloaded, loaded about 3x higher IIRC--part of that can surely be attributed to the messy breadboard layout).

Time to design a PCB--any hints for the placement and routing? Will become a dual layer board.

Samuel