Post by thread:[EE] Continuous high-side drive

Hi all, I have a circuit which needs to switch a 48V supply (which can rise as high as 80V). The load is ground referenced and can draw as much as 100 Amps. The switch, however, need not open while any more than 1 Amp is flowing. To accomplish this, I was planning on using several N channel FETs in parallel (because of their low Rds_on). Because the load can be on for hours or days, I need a true DC high-side drive (normal bootstrap-type drivers will not work because the bootstrap cap will discharge slowly). Based on a web search, I thought that the Intersil ISL6801 would do this, but the datasheet is confusing and ambiguous, and my actual prototype test showed that it cannot handle this situation (acts like a normal bootstrap driver). Most of the other ICs I found which claim to do this are limited to less than 80V at the FET source. I designed up a circuit to do it using a transformer and optocouplers, but it is not pretty or cheap. Anyone know of an IC to do gate drive for something like this? Thanks, Sean

Sean Breheny wrote: > I have a circuit which needs to switch a 48V supply (which can rise as > high as 80V). The load is ground referenced and can draw as much as > 100 Amps. The switch, however, need not open while any more than 1 Amp > is flowing. <snip> > Anyone know of an IC to do gate drive for something like this? Yes, The following are an absolutley brilliant invention and not well enough known (IMHO)... http://www.irf.com/product-info/datasheets/data/pvin.pdf Think of it as an optocoupler with an LED on one side and a solar cell on the other side. Output voltage is completely isolated. Limited in it's applications by it's output current and thereby turn on speed when switching MOSFET's. Potentialy (no pun intended) could be exactly what you need. -- Brent Brown, Electronic Design Solutions 16 English Street, St Andrews, Hamilton 3200, New Zealand Ph: +64 7 849 0069 Fax: +64 7 849 0071 Cell: 027 433 4069 eMail: spam_OUTbrent.brownTakeThisOuTclear.net.nz

Sean Breheny wrote: > Hi all, > > I have a circuit which needs to switch a 48V supply (which can rise as > high as 80V). The load is ground referenced and can draw as much as > 100 Amps. The switch, however, need not open while any more than 1 Amp > is flowing. > > To accomplish this, I was planning on using several N channel FETs in > parallel (because of their low Rds_on). Because the load can be on for > hours or days, I need a true DC high-side drive (normal bootstrap-type > drivers will not work because the bootstrap cap will discharge > slowly). Why not break your FETs into multiple banks and drive then with a multiple phase low off duty cycle signal? You could then use bootstrap drive mode. (clamp diode to give fast recharge recovery) If you have 8 Fets, you'd have 1 of 8 off for a few milliseconds every second so at most you'd need 1 extra Fet to cover the load of the one that is off.l > Based on a web search, I thought that the Intersil ISL6801 would do > this, but the datasheet is confusing and ambiguous, and my actual > prototype test showed that it cannot handle this situation (acts like > a normal bootstrap driver). Huh? The spec sheet says it does DC. "To ensure static DC operation an integrated recharge path charges the bootstrap cap while the driver is switched off." > Most of the other ICs I found which claim to do this are limited to > less than 80V at the FET source. > > I designed up a circuit to do it using a transformer and optocouplers, > but it is not pretty or cheap. > > Anyone know of an IC to do gate drive for something like this? And you can't use a relay because??? Or bridge the Fets with a relay after their initial operation? Robert

Brent Brown wrote: {Quote hidden}> Sean Breheny wrote: > > I have a circuit which needs to switch a 48V supply (which can rise > > as high as 80V). The load is ground referenced and can draw as much > > as 100 Amps. The switch, however, need not open while any more than > > 1 Amp is flowing. > <snip> > > Anyone know of an IC to do gate drive for something like this? > > Yes, > > The following are an absolutley brilliant invention and not well > enough known (IMHO)... > > www.irf.com/product-info/datasheets/data/pvin.pdf > > Think of it as an optocoupler with an LED on one side and a solar cell > on the other side. Output voltage is completely isolated. Limited in > it's applications by it's output current and thereby turn on speed > when switching MOSFET's. Potentialy (no pun intended) could be exactly > what you need. P.S. For more information check out their application note... http://www.irf.com/technical-info/appnotes/an-1017.pdf -- Brent Brown, Electronic Design Solutions 16 English Street, St Andrews, Hamilton 3200, New Zealand Ph: +64 7 849 0069 Fax: +64 7 849 0071 Cell: 027 433 4069 eMail: .....brent.brownKILLspam@spam@clear.net.nz

Sean Breheny wrote: {Quote hidden}> Hi all, > > I have a circuit which needs to switch a 48V supply (which can rise as > high as 80V). The load is ground referenced and can draw as much as > 100 Amps. The switch, however, need not open while any more than 1 Amp > is flowing. > > To accomplish this, I was planning on using several N channel FETs in > parallel (because of their low Rds_on). Because the load can be on for > hours or days, I need a true DC high-side drive (normal bootstrap-type > drivers will not work because the bootstrap cap will discharge > slowly). > > Based on a web search, I thought that the Intersil ISL6801 would do > this, but the datasheet is confusing and ambiguous, and my actual > prototype test showed that it cannot handle this situation (acts like > a normal bootstrap driver). > > Most of the other ICs I found which claim to do this are limited to > less than 80V at the FET source. > > I designed up a circuit to do it using a transformer and optocouplers, > but it is not pretty or cheap. > > Anyone know of an IC to do gate drive for something like this? > > Thanks, > > Sean > no but I would be interested in which ones you found for < 80v if you happen to have them handy?

part 1 1926 bytes content-type:text/plain; (unknown type 8bit not decoded) > I have a circuit which needs to switch a 48V supply (which can rise as > high as 80V). The load is ground referenced and can draw as much as > 100 Amps. The switch, however, need not open while any more than 1 Amp > is flowing. > > To accomplish this, I was planning on using several N channel FETs in > parallel (because of their low Rds_on). Because the load can be on for > hours or days, I need a true DC high-side drive (normal bootstrap-type > drivers will not work because the bootstrap cap will discharge > slowly). i think we can solve this one with a variation on the idea of a charge pump. http://img155.imageshack.us/img155/8071/drivecircuitei5.png when the square wave is low the pump capacitor charges from the "high" voltage supply when it is low its output is lifted and transferred into the tank capacitor. hence the tank capacitor output ends up at roughly the supply voltage for the high supply plus the peak to peak of the square wave minus a couple of diode drops and some other inefficiancies. the two diodes to the supplies for the square wave are to protect the square wave generator during power up and down and probablly need to be shotkeys. this then powers the pull up for the resistor/transistor buffer allowing it to get above the high voltage supply rail and hence provide and appropriate gate voltage for the main mosfets. working out component values and simulating this thing is left as an excercise for the implementer, this is an idea only and has not been tested. note: this circuit may have pathalogical behaviour if the supply voltage varies in step with the square wave driving the pump. This seems pretty unlikely thought -- No virus found in this outgoing message. Checked by AVG Free Edition. Version: 7.5.446 / Virus Database: 268.18.18/733 - Release Date: 25/03/2007 11:07 part 2 35 bytes content-type:text/plain; charset="us-ascii" (decoded 7bit)

Hi Sean, It's not as difficult with the normal bootstrap drivers as you might think. The High side drivers just need a supply referenced to the Source of the high side FET that allows them to drive with more than 10V. We did this for a military project using a 5V to 12V isolated DC to DC module powering an IR2104. The output ground of the modules are connected to the sources of the upper FET in the H-Bridge. The IR2104 switches this 12V onto the gate of the FET. The bootstrap method where the PWM provides the drive voltage is only a cost saving feature to provide a higher than motor rail voltage. You don't have to use it. John Dammeyer Automation Artisans Inc. http://www.autoartisans.com Ph. 1 250 544 4950 > {Original Message removed}

On 3/26/07, Sean Breheny <shb7KILLspamcornell.edu> wrote: > Hi all, > > I have a circuit which needs to switch a 48V supply (which can rise as > high as 80V). The load is ground referenced and can draw as much as > 100 Amps. The switch, however, need not open while any more than 1 Amp > is flowing. Telco/Central Office/Inside-plant remote power switches do stuff like this all the time. You might want to talk to or look at products like those... http://www.servertech.com would be one example of many... Many of these products do just have good "old-fashioned" relays in them, however... especially for the higher current models. Nate

Hi Brent, Thanks! This looks like exactly what I am looking for. Sean On 3/26/07, Brent Brown <.....brent.brownKILLspam.....clear.net.nz> wrote: {Quote hidden}> > Yes, > > > > The following are an absolutley brilliant invention and not well > > enough known (IMHO)... > > > > www.irf.com/product-info/datasheets/data/pvin.pdf > > > > Think of it as an optocoupler with an LED on one side and a solar cell > > on the other side. Output voltage is completely isolated. Limited in > > it's applications by it's output current and thereby turn on speed > > when switching MOSFET's. Potentialy (no pun intended) could be exactly > > what you need. > > P.S. For more information check out their application note... > > http://www.irf.com/technical-info/appnotes/an-1017.pdf > > > -- > Brent Brown, Electronic Design Solutions > 16 English Street, St Andrews, > Hamilton 3200, New Zealand > Ph: +64 7 849 0069 > Fax: +64 7 849 0071 > Cell: 027 433 4069 > eMail: EraseMEbrent.brownspam_OUTTakeThisOuTclear.net.nz > > > -

Hi John, Thanks for your reply. I was trying to do this without the cost of an isolated DC to DC converter (especially considering that only microamps are needed to keep the FET gates charged). Sean On 3/26/07, John Dammeyer <johndspam_OUTautoartisans.com> wrote: {Quote hidden}> Hi Sean, > > It's not as difficult with the normal bootstrap drivers as you might > think. > > The High side drivers just need a supply referenced to the Source of the > high side FET that allows them to drive with more than 10V. > > We did this for a military project using a 5V to 12V isolated DC to DC > module powering an IR2104. The output ground of the modules are > connected to the sources of the upper FET in the H-Bridge. The IR2104 > switches this 12V onto the gate of the FET. > > The bootstrap method where the PWM provides the drive voltage is only a > cost saving feature to provide a higher than motor rail voltage. You > don't have to use it. > > John Dammeyer > > > Automation Artisans Inc. > http://www.autoartisans.com > Ph. 1 250 544 4950 > > > > {Original Message removed}

Hi Robert, Thanks for your reply. Please see my responses below. On 3/26/07, Robert Rolf <@spam@Robert.RolfKILLspamualberta.ca> wrote: > Why not break your FETs into multiple banks and drive then with a > multiple phase low off duty cycle signal? > You could then use bootstrap drive mode. (clamp diode to give fast > recharge recovery) > If you have 8 Fets, you'd have 1 of 8 off for a few milliseconds > every second so at most you'd need 1 extra Fet to cover the load > of the one that is off.l That's not all that bad an idea but I had a feeling that there was a simpler way to do this which was a standard "easy way" to do this kind of thing. Brent seems to have such an answer. > Huh? The spec sheet says it does DC. > > "To ensure static DC operation an integrated recharge path charges the > bootstrap cap while the driver is switched off." > That's what I thought, too. However, read further and you will see that they are never clear about how that would work. There is no charge pump or other boosting supply present. Also, they talk about the bootstrap cap being charged "when the driver is off", which sounds suspiciously like a normal bootstrap driver. The "recharge path" they show would only help when the driver was in fact off. It shows no way of charging while the FET is on. My tests confirmed this - after turn on the gate drive slowly fell off until the source was lower than the gate drive bias supply. I think it must be poorly translated or misunderstood by the datasheet writer. I think what they actually mean is that it is like a normal bootstrapped driver except that they provide an internal path to recharge the cap so you do not need a low-side switch (which would normally ground the source node and charge the cap). > And you can't use a relay because??? > Or bridge the Fets with a relay after their initial operation? We looked at relays and couldn't find anything which could break 80VDC and also carry 100A while closed for anything less than $90 USD in large quantity. Most power relays are only spec'd to 32VDC. Sean > > Robert > > -

Hi Nate, Thanks. I think I'm going to try Brent's idea. Sean On 3/26/07, Nate Duehr <RemoveMEnateTakeThisOuTnatetech.com> wrote: {Quote hidden}> On 3/26/07, Sean Breheny <spamBeGoneshb7spamBeGonecornell.edu> wrote: > > Hi all, > > > > I have a circuit which needs to switch a 48V supply (which can rise as > > high as 80V). The load is ground referenced and can draw as much as > > 100 Amps. The switch, however, need not open while any more than 1 Amp > > is flowing. > > Telco/Central Office/Inside-plant remote power switches do stuff like > this all the time. > > You might want to talk to or look at products like those... > http://www.servertech.com would be one example of many... > > Many of these products do just have good "old-fashioned" relays in > them, however... especially for the higher current models. > > Nate > -

Hi Peter, I'm sure something like this would work but I'm going to try the integrated solution which Brent suggested first. Thanks! Sean On 3/26/07, peter green <TakeThisOuTplugwashEraseMEspam_OUTp10link.net> wrote: {Quote hidden}> i think we can solve this one with a variation on the idea of a charge pump. > > http://img155.imageshack.us/img155/8071/drivecircuitei5.png > > when the square wave is low the pump capacitor charges from the "high" voltage supply when it is low its output is lifted and transferred into the tank capacitor. > > hence the tank capacitor output ends up at roughly the supply voltage for the high supply plus the peak to peak of the square wave minus a couple of diode drops and some other inefficiancies. > > the two diodes to the supplies for the square wave are to protect the square wave generator during power up and down and probablly need to be shotkeys. > > this then powers the pull up for the resistor/transistor buffer allowing it to get above the high voltage supply rail and hence provide and appropriate gate voltage for the main mosfets. > > working out component values and simulating this thing is left as an excercise for the implementer, this is an idea only and has not been tested. > > note: this circuit may have pathalogical behaviour if the supply voltage varies in step with the square wave driving the pump. This seems pretty unlikely thought > -- > No virus found in this outgoing message. > Checked by AVG Free Edition. > Version: 7.5.446 / Virus Database: 268.18.18/733 - Release Date: 25/03/2007 11:07 > > > > -

On Mon, 2007-03-26 at 19:56 -0400, Sean Breheny wrote: > Hi all, > > I have a circuit which needs to switch a 48V supply (which can rise as > high as 80V). The load is ground referenced and can draw as much as > 100 Amps. The switch, however, need not open while any more than 1 Amp > is flowing. Have you considered using a relay? This it the typical type of application for a relay, unless speed or power consumption is really important. TTYL

> We looked at relays and couldn't find anything which could break 80VDC > and also carry 100A while closed for anything less than $90 USD in > large quantity. Most power relays are only spec'd to 32VDC. > > Sean > > > > > > Robert > > Robert, Check Allbright, Prestolite, AMTEK (spelling?). Also there are some Chinese contactors (Guizhou Tianyi Electrical) that may meet your price range. We use contactors up to about 1200A at similar voltages. 100A is at the small end of things! RP

Sean Breheny <shb7 <at> cornell.edu> writes: > > http://www.irf.com/technical-info/appnotes/an-1017.pdf Imho this is a great part but simulating the circuit first could be important. When I had to do such a floating switch I used a Royer style dc/dc converter and a hysteresis element after the high side rectifier. Imho the control current supplied by the irf part is insufficient for a stable gate drive when noise etc is present and during switching. Perhaps it could be used to supply power to a standard gate driver instead of being used stand-alone. Peter P.

Hi Richard, That was I who made the comment about not finding inexpensive relays, not Robert. I should have removed his name when I edited the quoted text. To really justify going to a relay, I'd say it would have to be less than $20 in high quantity. $90 is not my requirement, simply the cheapest I found. Sean On 3/26/07, Richard Prosser <RemoveMErhprosserTakeThisOuTgmail.com> wrote: {Quote hidden}> > We looked at relays and couldn't find anything which could break 80VDC > > and also carry 100A while closed for anything less than $90 USD in > > large quantity. Most power relays are only spec'd to 32VDC. > > > > Sean > > > > > > > > > > Robert > > > > > Robert, > Check Allbright, Prestolite, AMTEK (spelling?). Also there are some > Chinese contactors (Guizhou Tianyi Electrical) that may meet your > price range. > > We use contactors up to about 1200A at similar voltages. 100A is at > the small end of things! > > RP > -

Hi all, Thanks again for the help. I got some of the PVI1050N chips which Brent mentioned. They work great and look like an ideal solution for me. I don't care about turn-on time, only turn off time. I was able to speed that up by adding a BJT and a diode so that the PVI charges the FET gate through the diode and pulls it low using the transistor. This gave me 2ms turn-on time and 5us turn off time for 30nF gate capacitance (three FETs in parallel). Sean On 3/27/07, Sean Breheny <shb7EraseME.....cornell.edu> wrote: {Quote hidden}> Hi Richard, > > That was I who made the comment about not finding inexpensive relays, > not Robert. I should have removed his name when I edited the quoted > text. > > To really justify going to a relay, I'd say it would have to be less > than $20 in high quantity. $90 is not my requirement, simply the > cheapest I found. > > Sean > > On 3/26/07, Richard Prosser <EraseMErhprossergmail.com> wrote: > > > We looked at relays and couldn't find anything which could break 80VDC > > > and also carry 100A while closed for anything less than $90 USD in > > > large quantity. Most power relays are only spec'd to 32VDC. > > > > > > Sean > > > > > > > > > > > > > > Robert > > > > > > > > Robert, > > Check Allbright, Prestolite, AMTEK (spelling?). Also there are some > > Chinese contactors (Guizhou Tianyi Electrical) that may meet your > > price range. > > > > We use contactors up to about 1200A at similar voltages. 100A is at > > the small end of things! > > > > RP > > --