EDAboard.com | EDAboard.de | EDAboard.co.uk | WTWH Media

H-Bridge with level shift - MOSFET gate voltage issues?

Ask a question - edaboard.com

elektroda.net NewsGroups Forum Index - Electronics Design - H-Bridge with level shift - MOSFET gate voltage issues?

Goto page Previous  1, 2, 3, 4, 5  Next

John Larkin
Guest

Fri Jan 04, 2019 12:45 am   



On Thu, 03 Jan 2019 14:12:23 -0600, Joe Chisolm
<jchisolm6_at_earthlink.net> wrote:

Quote:
On Thu, 03 Jan 2019 07:16:36 -0800, John Larkin wrote:

On Wed, 2 Jan 2019 08:46:59 +0000, Tim Watts <tw_at_example.com> wrote:

Hi,

Not much good with analogue, would be grateful for any comments on this
please :)

https://www.circuitlab.com/circuit/a5399q386x6h/test-h-bridge/

(or simple picture) https://ibb.co/vh7GRG2

That's half an H-Bridge, we'll have 2 in full H-Bridge configuration
driving a 30V LED garden light string from a Raspberry Pi (details
below). R2/R3 are dummy loads for simulation, please ignore.

Problems:
=========
A) It's possible I might be exceeding gate-source voltages on TR2 or TR4
(TR4 max g-d V is 20V). Is there a standard solution to this? Something
with a zener maybe?


B) Transistor selection is not final - and TR1 may not work with 3.3V
input (I designed to 5V initially) so we may need a different device or
a buffer gate. I can solve that one way or another - just noting I am
aware of this :)


Many thanks,

Tim

What and why:
=============

Application is Raspberry Pi GPIO (3.3V) to 30V LED lighting string.
There'll be 2 of these in full H-Bridge configuration as the LED string
is a mixture of forward and reverse connected substrings. Son wants to
make the patterns more interesting and likes Pi's so it's a fun project.

What I've done so far:

1) Not driving an inductive load, so lose the flywheel diodes common in
motor drives;

2) I've tried it in CircuitLab with and without the TR2/3 driver stage.
The switching speed and power dissipation is a lot better with that
stage at higher switching rates. Whilst we probably won't drive this
much over 1kHz, I prefer to not have software errors able to overheat
things and we need small heatsinks as this is going in a waterproof box
on the lawn;

3) Ignore R2/R3 - this is a dummy load for CircuitLab simulation;

4) R4 is a dummy internal resistance for the PSU (otherwise we get
impossibly high switching transients). In reality the 30V supply is a
current limited switch mode SELV** PSU.

** Separated Extra Low Voltage - UK designation for a "safe to touch"
PSU for wet applications

Why an h-bridge? You should be able to switch/dim a string of LEDs
with a single logic-level-gate mosfet.

The FDC5614P is a logic level P. 60V, 3A, Vgs -1.6
NTF3055L108 - logic level N channel 60V 3 A
Drive them from gpio pins on the Pi.
If you decide to not roll your own, look at the DRV8871 from TI

As, I think JL mentioned, use AC supply and a couple of SCRs with a
classic dimmer control. You would need zero crossing detector for the pi.


AC and a triac, with some clever gate timing, could do the same
lighting tricks as two half-bridges.


--

John Larkin Highland Technology, Inc
picosecond timing precision measurement

jlarkin att highlandtechnology dott com
http://www.highlandtechnology.com

Lasse Langwadt Christense
Guest

Fri Jan 04, 2019 12:45 am   



fredag den 4. januar 2019 kl. 00.03.40 UTC+1 skrev John Larkin:
Quote:
On Thu, 03 Jan 2019 14:12:23 -0600, Joe Chisolm
jchisolm6_at_earthlink.net> wrote:

On Thu, 03 Jan 2019 07:16:36 -0800, John Larkin wrote:

On Wed, 2 Jan 2019 08:46:59 +0000, Tim Watts <tw_at_example.com> wrote:

Hi,

Not much good with analogue, would be grateful for any comments on this
please :)

https://www.circuitlab.com/circuit/a5399q386x6h/test-h-bridge/

(or simple picture) https://ibb.co/vh7GRG2

That's half an H-Bridge, we'll have 2 in full H-Bridge configuration
driving a 30V LED garden light string from a Raspberry Pi (details
below). R2/R3 are dummy loads for simulation, please ignore.

Problems:
=========
A) It's possible I might be exceeding gate-source voltages on TR2 or TR4
(TR4 max g-d V is 20V). Is there a standard solution to this? Something
with a zener maybe?


B) Transistor selection is not final - and TR1 may not work with 3.3V
input (I designed to 5V initially) so we may need a different device or
a buffer gate. I can solve that one way or another - just noting I am
aware of this :)


Many thanks,

Tim

What and why:
=============

Application is Raspberry Pi GPIO (3.3V) to 30V LED lighting string.
There'll be 2 of these in full H-Bridge configuration as the LED string
is a mixture of forward and reverse connected substrings. Son wants to
make the patterns more interesting and likes Pi's so it's a fun project.

What I've done so far:

1) Not driving an inductive load, so lose the flywheel diodes common in
motor drives;

2) I've tried it in CircuitLab with and without the TR2/3 driver stage.
The switching speed and power dissipation is a lot better with that
stage at higher switching rates. Whilst we probably won't drive this
much over 1kHz, I prefer to not have software errors able to overheat
things and we need small heatsinks as this is going in a waterproof box
on the lawn;

3) Ignore R2/R3 - this is a dummy load for CircuitLab simulation;

4) R4 is a dummy internal resistance for the PSU (otherwise we get
impossibly high switching transients). In reality the 30V supply is a
current limited switch mode SELV** PSU.

** Separated Extra Low Voltage - UK designation for a "safe to touch"
PSU for wet applications

Why an h-bridge? You should be able to switch/dim a string of LEDs
with a single logic-level-gate mosfet.

The FDC5614P is a logic level P. 60V, 3A, Vgs -1.6
NTF3055L108 - logic level N channel 60V 3 A
Drive them from gpio pins on the Pi.
If you decide to not roll your own, look at the DRV8871 from TI

As, I think JL mentioned, use AC supply and a couple of SCRs with a
classic dimmer control. You would need zero crossing detector for the pi.

AC and a triac, with some clever gate timing, could do the same
lighting tricks as two half-bridges.


but then you have to find a real transformer instead of a cheap switching supply

John Larkin
Guest

Fri Jan 04, 2019 12:45 am   



On Thu, 3 Jan 2019 22:24:24 +0000, Tim Watts <tw_at_example.com> wrote:

Quote:
On 03/01/2019 18:36, John Larkin wrote:

OK, you can do tricks if you drive the half-bridges indepently.

Ha - sorry, I never occurred to me to actually point that out :-|

I do apologise for any confusion there...


I can generate my own confusion without help from anyone.


--

John Larkin Highland Technology, Inc
picosecond timing precision measurement

jlarkin att highlandtechnology dott com
http://www.highlandtechnology.com


Guest

Fri Jan 04, 2019 2:45 am   



On Friday, January 4, 2019 at 6:27:03 AM UTC+11, mako...@yahoo.com wrote:
Quote:
have you looked at cross conduction?

as the gate voltage moves between its upper and lower values, is there a time, even a brief time, when both the upper and lower fets are both on.?


My evil high voltage driver had lumped constant delay line so you didn't start turning on the upper fet until the lower fet was off, and vice versa.

74121/74123 were too slow, and the sort of ECL parts that were fast enough were even more expensive (and obsolete now). I've done discrete emitter-coupled monostables that would have been fast enough, but the board had to be compact.

> If so, you need to include current limiting resistors.

The current surge is brief - and since the devices take a while to turn fully on - not all that big. But big enough and fast enough to be an interference problem.

> or add level shifters such that there is a time when both upper and lower FETs are off and there is no cross conduction.

It's a timing problem. If your level shifters provided the right sort of delays, they could work.

> I prefer belts and suspenders and would do both.

The absolute maximum gate-to-source voltage ratings mean that you do need level shifters. You also have to get the timing right, so you are probably stuck with doing both.

--
Bill Sloman, Sydney

Jasen Betts
Guest

Fri Jan 04, 2019 4:45 am   



On 2019-01-03, Joe Chisolm <jchisolm6_at_earthlink.net> wrote:
Quote:
On Thu, 03 Jan 2019 07:16:36 -0800, John Larkin wrote:

On Wed, 2 Jan 2019 08:46:59 +0000, Tim Watts <tw_at_example.com> wrote:

Hi,

Not much good with analogue, would be grateful for any comments on this
please :)

https://www.circuitlab.com/circuit/a5399q386x6h/test-h-bridge/

(or simple picture) https://ibb.co/vh7GRG2

That's half an H-Bridge, we'll have 2 in full H-Bridge configuration
driving a 30V LED garden light string from a Raspberry Pi (details
below). R2/R3 are dummy loads for simulation, please ignore.

Problems:
=========
A) It's possible I might be exceeding gate-source voltages on TR2 or TR4
(TR4 max g-d V is 20V). Is there a standard solution to this? Something
with a zener maybe?


B) Transistor selection is not final - and TR1 may not work with 3.3V
input (I designed to 5V initially) so we may need a different device or
a buffer gate. I can solve that one way or another - just noting I am
aware of this :)


Many thanks,

Tim

What and why:
=============

Application is Raspberry Pi GPIO (3.3V) to 30V LED lighting string.
There'll be 2 of these in full H-Bridge configuration as the LED string
is a mixture of forward and reverse connected substrings. Son wants to
make the patterns more interesting and likes Pi's so it's a fun project.

What I've done so far:

1) Not driving an inductive load, so lose the flywheel diodes common in
motor drives;

2) I've tried it in CircuitLab with and without the TR2/3 driver stage.
The switching speed and power dissipation is a lot better with that
stage at higher switching rates. Whilst we probably won't drive this
much over 1kHz, I prefer to not have software errors able to overheat
things and we need small heatsinks as this is going in a waterproof box
on the lawn;

3) Ignore R2/R3 - this is a dummy load for CircuitLab simulation;

4) R4 is a dummy internal resistance for the PSU (otherwise we get
impossibly high switching transients). In reality the 30V supply is a
current limited switch mode SELV** PSU.

** Separated Extra Low Voltage - UK designation for a "safe to touch"
PSU for wet applications

Why an h-bridge? You should be able to switch/dim a string of LEDs
with a single logic-level-gate mosfet.

The FDC5614P is a logic level P. 60V, 3A, Vgs -1.6
NTF3055L108 - logic level N channel 60V 3 A
Drive them from gpio pins on the Pi.
If you decide to not roll your own, look at the DRV8871 from TI

As, I think JL mentioned, use AC supply and a couple of SCRs with a
classic dimmer control. You would need zero crossing detector for the pi.


no, you'd need digital pots and a phase-shift network, Pi (running
windows or linux) is not realtime enough to chase zero crossings.

--
When I tried casting out nines I made a hash of it.

Klaus Kragelund
Guest

Fri Jan 04, 2019 6:45 am   



Check out Tim Williams site, he has different gatedriver designs if you really want to go down that troublesome road

He is a regular here

Cheers

Klaus

Tim Watts
Guest

Fri Jan 04, 2019 10:45 am   



On 04/01/2019 05:29, Klaus Kragelund wrote:
Quote:
Check out Tim Williams site, he has different gatedriver designs if you really want to go down that troublesome road

He is a regular here

Cheers

Klaus


Thanks - I will look him up on Google :)

--
Email does not work

Klaus Kragelund
Guest

Fri Jan 04, 2019 10:45 am   



Tims site , direct link to gatedriver:

https://www.seventransistorlabs.com/ClassD1/index.html#gd


Guest

Fri Jan 04, 2019 7:45 pm   



On Wednesday, January 2, 2019 at 3:47:07 AM UTC-5, Tim Watts wrote:
Quote:
Hi,

Not much good with analogue, would be grateful for any comments on this
please :)

https://www.circuitlab.com/circuit/a5399q386x6h/test-h-bridge/

(or simple picture) https://ibb.co/vh7GRG2

That's half an H-Bridge, we'll have 2 in full H-Bridge configuration
driving a 30V LED garden light string from a Raspberry Pi (details
below). R2/R3 are dummy loads for simulation, please ignore.

Problems:
=========
A) It's possible I might be exceeding gate-source voltages on TR2 or TR4
(TR4 max g-d V is 20V). Is there a standard solution to this? Something
with a zener maybe?


B) Transistor selection is not final - and TR1 may not work with 3.3V
input (I designed to 5V initially) so we may need a different device or
a buffer gate. I can solve that one way or another - just noting I am
aware of this :)


Many thanks,

Tim

What and why:
=============

Application is Raspberry Pi GPIO (3.3V) to 30V LED lighting string.
There'll be 2 of these in full H-Bridge configuration as the LED string
is a mixture of forward and reverse connected substrings. Son wants to
make the patterns more interesting and likes Pi's so it's a fun project.

What I've done so far:

1) Not driving an inductive load, so lose the flywheel diodes common in
motor drives;

2) I've tried it in CircuitLab with and without the TR2/3 driver stage.
The switching speed and power dissipation is a lot better with that
stage at higher switching rates. Whilst we probably won't drive this
much over 1kHz, I prefer to not have software errors able to overheat
things and we need small heatsinks as this is going in a waterproof box
on the lawn;

3) Ignore R2/R3 - this is a dummy load for CircuitLab simulation;

4) R4 is a dummy internal resistance for the PSU (otherwise we get
impossibly high switching transients). In reality the 30V supply is a
current limited switch mode SELV** PSU.

** Separated Extra Low Voltage - UK designation for a "safe to touch"
PSU for wet applications



--
Email does not work


You're building a neighborhood EMI generator. Long line lengths make good radiaors. If you can't fix that, don't go any further.

John Larkin
Guest

Fri Jan 04, 2019 8:45 pm   



On Fri, 4 Jan 2019 10:29:01 -0800 (PST),
bloggs.fredbloggs.fred_at_gmail.com wrote:

Quote:
On Wednesday, January 2, 2019 at 3:47:07 AM UTC-5, Tim Watts wrote:
Hi,

Not much good with analogue, would be grateful for any comments on this
please :)

https://www.circuitlab.com/circuit/a5399q386x6h/test-h-bridge/

(or simple picture) https://ibb.co/vh7GRG2

That's half an H-Bridge, we'll have 2 in full H-Bridge configuration
driving a 30V LED garden light string from a Raspberry Pi (details
below). R2/R3 are dummy loads for simulation, please ignore.

Problems:
=========
A) It's possible I might be exceeding gate-source voltages on TR2 or TR4
(TR4 max g-d V is 20V). Is there a standard solution to this? Something
with a zener maybe?


B) Transistor selection is not final - and TR1 may not work with 3.3V
input (I designed to 5V initially) so we may need a different device or
a buffer gate. I can solve that one way or another - just noting I am
aware of this :)


Many thanks,

Tim

What and why:
=============

Application is Raspberry Pi GPIO (3.3V) to 30V LED lighting string.
There'll be 2 of these in full H-Bridge configuration as the LED string
is a mixture of forward and reverse connected substrings. Son wants to
make the patterns more interesting and likes Pi's so it's a fun project.

What I've done so far:

1) Not driving an inductive load, so lose the flywheel diodes common in
motor drives;

2) I've tried it in CircuitLab with and without the TR2/3 driver stage.
The switching speed and power dissipation is a lot better with that
stage at higher switching rates. Whilst we probably won't drive this
much over 1kHz, I prefer to not have software errors able to overheat
things and we need small heatsinks as this is going in a waterproof box
on the lawn;

3) Ignore R2/R3 - this is a dummy load for CircuitLab simulation;

4) R4 is a dummy internal resistance for the PSU (otherwise we get
impossibly high switching transients). In reality the 30V supply is a
current limited switch mode SELV** PSU.

** Separated Extra Low Voltage - UK designation for a "safe to touch"
PSU for wet applications



--
Email does not work

You're building a neighborhood EMI generator. Long line lengths make good radiaors. If you can't fix that, don't go any further.


Switching at KHz, nobody would notice.

I have an Amazon stick-on dimmable LED strip for by workbench. It PWMs
at about 4 KHz.





--

John Larkin Highland Technology, Inc
picosecond timing precision measurement

jlarkin att highlandtechnology dott com
http://www.highlandtechnology.com


Guest

Fri Jan 04, 2019 10:45 pm   



On Friday, January 4, 2019 at 2:34:50 PM UTC-5, John Larkin wrote:
Quote:
On Fri, 4 Jan 2019 10:29:01 -0800 (PST),
bloggs.fredbloggs.fred_at_gmail.com wrote:

On Wednesday, January 2, 2019 at 3:47:07 AM UTC-5, Tim Watts wrote:
Hi,

Not much good with analogue, would be grateful for any comments on this
please :)

https://www.circuitlab.com/circuit/a5399q386x6h/test-h-bridge/

(or simple picture) https://ibb.co/vh7GRG2

That's half an H-Bridge, we'll have 2 in full H-Bridge configuration
driving a 30V LED garden light string from a Raspberry Pi (details
below). R2/R3 are dummy loads for simulation, please ignore.

Problems:
=========
A) It's possible I might be exceeding gate-source voltages on TR2 or TR4
(TR4 max g-d V is 20V). Is there a standard solution to this? Something
with a zener maybe?


B) Transistor selection is not final - and TR1 may not work with 3.3V
input (I designed to 5V initially) so we may need a different device or
a buffer gate. I can solve that one way or another - just noting I am
aware of this :)


Many thanks,

Tim

What and why:
=============

Application is Raspberry Pi GPIO (3.3V) to 30V LED lighting string.
There'll be 2 of these in full H-Bridge configuration as the LED string
is a mixture of forward and reverse connected substrings. Son wants to
make the patterns more interesting and likes Pi's so it's a fun project.

What I've done so far:

1) Not driving an inductive load, so lose the flywheel diodes common in
motor drives;

2) I've tried it in CircuitLab with and without the TR2/3 driver stage.
The switching speed and power dissipation is a lot better with that
stage at higher switching rates. Whilst we probably won't drive this
much over 1kHz, I prefer to not have software errors able to overheat
things and we need small heatsinks as this is going in a waterproof box
on the lawn;

3) Ignore R2/R3 - this is a dummy load for CircuitLab simulation;

4) R4 is a dummy internal resistance for the PSU (otherwise we get
impossibly high switching transients). In reality the 30V supply is a
current limited switch mode SELV** PSU.

** Separated Extra Low Voltage - UK designation for a "safe to touch"
PSU for wet applications



--
Email does not work

You're building a neighborhood EMI generator. Long line lengths make good radiaors. If you can't fix that, don't go any further.

Switching at KHz, nobody would notice.

I have an Amazon stick-on dimmable LED strip for by workbench. It PWMs
at about 4 KHz.


It is probably filtered.


Quote:





--

John Larkin Highland Technology, Inc
picosecond timing precision measurement

jlarkin att highlandtechnology dott com
http://www.highlandtechnology.com


Tim Watts
Guest

Sat Jan 05, 2019 11:45 am   



On 04/01/2019 18:29, bloggs.fredbloggs.fred_at_gmail.com wrote:

Quote:
You're building a neighborhood EMI generator. Long line lengths make good radiaors. If you can't fix that, don't go any further.


It's a good point -

I don't have access to a 'scope but would this not apply to most fairly
lights with dimming patterns?

My set cannot be using current control as you can odd anywhere from one
to 8 sets in parallel after the controller.

I find it doubtful they could be using voltage control given LEDs - or
could they?

These are definitely not addressable NeoPixel devices.

--
Email does not work

Tim Watts
Guest

Sat Jan 05, 2019 11:45 am   



On 04/01/2019 19:34, John Larkin wrote:

Quote:
I have an Amazon stick-on dimmable LED strip for by workbench. It PWMs
at about 4 KHz.


It was a good point to raise - to not try to get to clever and bang the
PWM through the roof for the sake of flicker control :)

In reality, we wouldn't want to try to get too high with a Pi, but kHz
sounds like a good place to be for LEDs.

--
Email does not work

Tim Watts
Guest

Sat Jan 05, 2019 11:45 am   



On 04/01/2019 20:52, bloggs.fredbloggs.fred_at_gmail.com wrote:

Quote:
I have an Amazon stick-on dimmable LED strip for by workbench. It PWMs
at about 4 KHz.

It is probably filtered.


I've wanted an excuse to get a picoscope or similar. Apart from helping
with software debugging, it would be interesting to see what the current
(approved) unit puts out.

As long as I stay in the same region, should be fine - there are no
black boxes on the strings that could contain much in the way of
filtering (or if there are chokes or caps in the LED bases, they would
work as well with my controller if the drive pattern is similar)


--
Email does not work

John Larkin
Guest

Sat Jan 05, 2019 5:45 pm   



On Sat, 5 Jan 2019 10:03:39 +0000, Tim Watts <tw_at_example.com> wrote:

Quote:
On 04/01/2019 19:34, John Larkin wrote:

I have an Amazon stick-on dimmable LED strip for by workbench. It PWMs
at about 4 KHz.

It was a good point to raise - to not try to get to clever and bang the
PWM through the roof for the sake of flicker control :)

In reality, we wouldn't want to try to get too high with a Pi, but kHz
sounds like a good place to be for LEDs.


If you switch at, say, 1 KHz, the radiation efficiency of a string of
LEDs as an antenna will be zilch. At the AM broadcast band, where the
antanna efficiency works its way up to terrible, you'd be looking at
the 1000th harmonic of the drive waveform. That line will contain
microscopic energy.

So don't worry about it.

But for extra fun, go software spread-spectrum.


--

John Larkin Highland Technology, Inc

lunatic fringe electronics

Goto page Previous  1, 2, 3, 4, 5  Next

elektroda.net NewsGroups Forum Index - Electronics Design - H-Bridge with level shift - MOSFET gate voltage issues?

Ask a question - edaboard.com

Arabic version Bulgarian version Catalan version Czech version Danish version German version Greek version English version Spanish version Finnish version French version Hindi version Croatian version Indonesian version Italian version Hebrew version Japanese version Korean version Lithuanian version Latvian version Dutch version Norwegian version Polish version Portuguese version Romanian version Russian version Slovak version Slovenian version Serbian version Swedish version Tagalog version Ukrainian version Vietnamese version Chinese version Turkish version
EDAboard.com map