GDT wire
zzz_julian_zzz, Tue Jun 30 2015, 03:04PMHi Everyone,
I'm convinced that in order to make a good GDT for coil, one of the criteria is to have a correct core, but what other than this??
In my test, wire being used has very huge effect on the leakage inductance / ringing on the GATE - EMMITER signal.
with UTP cable (awg# 24) leakage is still a bit high and causes high tops (38volts secondary on a 1:1 GDT with primary swing of 24v[N30 core]) .. Not all of this can be answered by just dampening the signal on a resistor on the gate. I tried using insulated magnet wire also awg 24 thick, and surprisingly the leakage L went down and the signal are much better (the pic with signal with low tops is the magnet wire)
I tried breaking the insulation of this MAGNET wire by supplying it with a fused 400vdc power. I twisted it like in GDT fashion and sourced it with 400vdc for about a minute. No shortage/flashover occur. My question is, is this 400vdc I'm planning to apply on a bridge will be the one i should be concerned about in terms of wire insulation? because the GDT might see more than 400vdc here since it is introduced in a AC inverter circuit that outputs +/-400vAC.
Can anyone enlighten me here? Thanks!
Re: GDT wire
loneoceans, Tue Jun 30 2015, 06:11PM
That’s correct, it's clear that reducing the thickness of the insulation of the wires will reduce the leakage inductance, and you can reduce it even further by using shielded twisted pair. Most commercial pulse transformers are specifically designed with thin insulation to reduce leakage inductance. However I think in practice, the usual twisted pair wires using Cat5 cable works fine.
In fact your GDT output on the gates looks fine (even with your conventional UTP cable). I would recommend using 30V TVSes to avoid damaging your gate insulation.
As for magnet wire insulation, it depends a lot of the wire insulation material and thickness. See this page here for detailed information on calculating the breakdown voltage (Ds). If you’re running say a 600V bus, most good quality magnet wire should be ok.
Depending on what GDT core you use, some of them might have hard edges (e.g. rectangular cross section toroids) - you'll need to be careful not to scrape the insulation off the magnet wire in those cases. :)
loneoceans, Tue Jun 30 2015, 06:11PM
zzz_julian_zzz wrote ...
Hi Everyone,
I'm convinced that in order to make a good GDT for coil, one of the criteria is to have a correct core, but what other than this??
In my test, wire being used has very huge effect on the leakage inductance / ringing on the GATE - EMMITER signal.
with UTP cable (awg# 24) leakage is still a bit high and causes high tops (38volts secondary on a 1:1 GDT with primary swing of 24v[N30 core]) .. Not all of this can be answered by just dampening the signal on a resistor on the gate. I tried using insulated magnet wire also awg 24 thick, and surprisingly the leakage L went down and the signal are much better (the pic with signal with low tops is the magnet wire)
I tried breaking the insulation of this MAGNET wire by supplying it with a fused 400vdc power. I twisted it like in GDT fashion and sourced it with 400vdc for about a minute. No shortage/flashover occur. My question is, is this 400vdc I'm planning to apply on a bridge will be the one i should be concerned about in terms of wire insulation? because the GDT might see more than 400vdc here since it is introduced in a AC inverter circuit that outputs +/-400vAC.
Can anyone enlighten me here? Thanks!
Hi Everyone,
I'm convinced that in order to make a good GDT for coil, one of the criteria is to have a correct core, but what other than this??
In my test, wire being used has very huge effect on the leakage inductance / ringing on the GATE - EMMITER signal.
with UTP cable (awg# 24) leakage is still a bit high and causes high tops (38volts secondary on a 1:1 GDT with primary swing of 24v[N30 core]) .. Not all of this can be answered by just dampening the signal on a resistor on the gate. I tried using insulated magnet wire also awg 24 thick, and surprisingly the leakage L went down and the signal are much better (the pic with signal with low tops is the magnet wire)
I tried breaking the insulation of this MAGNET wire by supplying it with a fused 400vdc power. I twisted it like in GDT fashion and sourced it with 400vdc for about a minute. No shortage/flashover occur. My question is, is this 400vdc I'm planning to apply on a bridge will be the one i should be concerned about in terms of wire insulation? because the GDT might see more than 400vdc here since it is introduced in a AC inverter circuit that outputs +/-400vAC.
Can anyone enlighten me here? Thanks!
That’s correct, it's clear that reducing the thickness of the insulation of the wires will reduce the leakage inductance, and you can reduce it even further by using shielded twisted pair. Most commercial pulse transformers are specifically designed with thin insulation to reduce leakage inductance. However I think in practice, the usual twisted pair wires using Cat5 cable works fine.
In fact your GDT output on the gates looks fine (even with your conventional UTP cable). I would recommend using 30V TVSes to avoid damaging your gate insulation.
As for magnet wire insulation, it depends a lot of the wire insulation material and thickness. See this page here for detailed information on calculating the breakdown voltage (Ds). If you’re running say a 600V bus, most good quality magnet wire should be ok.
Depending on what GDT core you use, some of them might have hard edges (e.g. rectangular cross section toroids) - you'll need to be careful not to scrape the insulation off the magnet wire in those cases. :)
Re: GDT wire
Hydron, Tue Jun 30 2015, 06:21PM
Kizmo (and I'm assuming others too) has used co-ax to make super low leakage inductance GDTs - using the core as primary and shield as secondary or vice versa.
He's got some pictures floating around of the setup (possibly on the forum), worth looking/asking if you're interested.
Hydron, Tue Jun 30 2015, 06:21PM
Kizmo (and I'm assuming others too) has used co-ax to make super low leakage inductance GDTs - using the core as primary and shield as secondary or vice versa.
He's got some pictures floating around of the setup (possibly on the forum), worth looking/asking if you're interested.
Re: GDT wire
Mads Barnkob, Tue Jun 30 2015, 08:49PM
Are your familiar with the tests thedatastream made long ago?
Mads Barnkob, Tue Jun 30 2015, 08:49PM
Are your familiar with the tests thedatastream made long ago?
Re: GDT wire
..., Tue Jun 30 2015, 09:21PM
I would shy away from magnet wire, the single GDT I wound with magnet wire was fine for many hours but eventually there was enough physical stress placed on the leads (due to moving the coil, etc) that it arced over and blew up a quite pricey transistor. Maybe if you pot the sucker in epoxy like a commercial transformer, otherwise I would recommend kynar insulated wire (ex, surplus wirewrap wire) if you really need the improved performance, or normal UTP for general work.
..., Tue Jun 30 2015, 09:21PM
I would shy away from magnet wire, the single GDT I wound with magnet wire was fine for many hours but eventually there was enough physical stress placed on the leads (due to moving the coil, etc) that it arced over and blew up a quite pricey transistor. Maybe if you pot the sucker in epoxy like a commercial transformer, otherwise I would recommend kynar insulated wire (ex, surplus wirewrap wire) if you really need the improved performance, or normal UTP for general work.
Re: GDT wire
klugesmith, Wed Jul 01 2015, 04:45AM
Today in a coilgun thread, Der Albi pointed out a cringeworthy video of magnet wire physical abuse.
Back to GDT's, has anyone tried bonded bifilar wire? has a representative picture:
I have a spool of untwisted bifilar magnet wire somewhere, with two contrasting insulation colors and a 50 ohm characteristic impedance. Traditional impedance value for twisted pairs of "wire wrap wire" is more like 110 ohms. It matter in baluns; I don't know about GDT's.
klugesmith, Wed Jul 01 2015, 04:45AM
Today in a coilgun thread, Der Albi pointed out a cringeworthy video of magnet wire physical abuse.
Back to GDT's, has anyone tried bonded bifilar wire?
has a representative picture:
I have a spool of untwisted bifilar magnet wire somewhere, with two contrasting insulation colors and a 50 ohm characteristic impedance. Traditional impedance value for twisted pairs of "wire wrap wire" is more like 110 ohms. It matter in baluns; I don't know about GDT's.
Re: GDT wire
zzz_julian_zzz, Thu Jul 02 2015, 02:22AM
Hi Loneoceans,
Thanks for the info. Yup, I'm using a smooth GDT core BUT, I think I don't want to try this wires anymore because of the insulation issues, and at stake of costly IGBTs :) I decided to just leave it with the trusty UTP since you said waveforms are ok in shape.
Thanks for your reply. Yup, I heard that co-ax cables are good for GDT since these are shielded, but honestly I didn't see anything like it before. I'll google for it. Thanks for the advice.
Hi Mads,
Yup, I saw that one before, unfortunately, that site didn't mentioned anything with regards to magnitude of wire insulation. The topic is mainly how to reduce Leakage L I think.. Thanks!
Hi ...,
Even before I would go into these problems of scratching the insulation sheets of magnetic wires, that sheet itself is my question. So, thanks for the advice, I would go for UTP...
Wow! Those wires have some extreme insulation and that guy on the vid is !
but I can't put those on a 1" size GDT core :) maybe the thinner ones :) thanks for the comment.
zzz_julian_zzz, Thu Jul 02 2015, 02:22AM
loneoceans wrote ...
fr LONEOCEANS,
That’s correct, it's clear that reducing the thickness of the insulation of the wires will reduce the leakage inductance, and you can reduce it even further by using shielded twisted pair. Most commercial pulse transformers are specifically designed with thin insulation to reduce leakage inductance. However I think in practice, the usual twisted pair wires using Cat5 cable works fine.
In fact your GDT output on the gates looks fine (even with your conventional UTP cable). I would recommend using 30V TVSes to avoid damaging your gate insulation.
As for magnet wire insulation, it depends a lot of the wire insulation material and thickness. See this page here for detailed information on calculating the breakdown voltage (Ds). If you’re running say a 600V bus, most good quality magnet wire should be ok.
Depending on what GDT core you use, some of them might have hard edges (e.g. rectangular cross section toroids) - you'll need to be careful not to scrape the insulation off the magnet wire in those cases. :)
fr LONEOCEANS,
That’s correct, it's clear that reducing the thickness of the insulation of the wires will reduce the leakage inductance, and you can reduce it even further by using shielded twisted pair. Most commercial pulse transformers are specifically designed with thin insulation to reduce leakage inductance. However I think in practice, the usual twisted pair wires using Cat5 cable works fine.
In fact your GDT output on the gates looks fine (even with your conventional UTP cable). I would recommend using 30V TVSes to avoid damaging your gate insulation.
As for magnet wire insulation, it depends a lot of the wire insulation material and thickness. See this page here for detailed information on calculating the breakdown voltage (Ds). If you’re running say a 600V bus, most good quality magnet wire should be ok.
Depending on what GDT core you use, some of them might have hard edges (e.g. rectangular cross section toroids) - you'll need to be careful not to scrape the insulation off the magnet wire in those cases. :)
Hi Loneoceans,
Thanks for the info. Yup, I'm using a smooth GDT core BUT, I think I don't want to try this wires anymore because of the insulation issues, and at stake of costly IGBTs :) I decided to just leave it with the trusty UTP since you said waveforms are ok in shape.
Hydron wrote ...
Kizmo (and I'm assuming others too) has used co-ax to make super low leakage inductance GDTs - using the core as primary and shield as secondary or vice versa.
He's got some pictures floating around of the setup (possibly on the forum), worth looking/asking if you're interested.
Hi Hydron,,Kizmo (and I'm assuming others too) has used co-ax to make super low leakage inductance GDTs - using the core as primary and shield as secondary or vice versa.
He's got some pictures floating around of the setup (possibly on the forum), worth looking/asking if you're interested.
Thanks for your reply. Yup, I heard that co-ax cables are good for GDT since these are shielded, but honestly I didn't see anything like it before. I'll google for it. Thanks for the advice.
Mads Barnkob wrote ...
Are your familiar with the tests thedatastream made long ago?
Are your familiar with the tests thedatastream made long ago?
Hi Mads,
Yup, I saw that one before, unfortunately, that site didn't mentioned anything with regards to magnitude of wire insulation. The topic is mainly how to reduce Leakage L I think.. Thanks!
... wrote ...
I would shy away from magnet wire, the single GDT I wound with magnet wire was fine for many hours but eventually there was enough physical stress placed on the leads (due to moving the coil, etc) that it arced over and blew up a quite pricey transistor. Maybe if you pot the sucker in epoxy like a commercial transformer, otherwise I would recommend kynar insulated wire (ex, surplus wirewrap wire) if you really need the improved performance, or normal UTP for general work.
I would shy away from magnet wire, the single GDT I wound with magnet wire was fine for many hours but eventually there was enough physical stress placed on the leads (due to moving the coil, etc) that it arced over and blew up a quite pricey transistor. Maybe if you pot the sucker in epoxy like a commercial transformer, otherwise I would recommend kynar insulated wire (ex, surplus wirewrap wire) if you really need the improved performance, or normal UTP for general work.
Hi ...,
Even before I would go into these problems of scratching the insulation sheets of magnetic wires, that sheet itself is my question. So, thanks for the advice, I would go for UTP...
klugesmith wrote ...
Today in a coilgun thread, Der Albi pointed out a cringeworthy video of magnet wire physical abuse.
Back to GDT's, has anyone tried bonded bifilar wire? has a representative picture:
I have a spool of untwisted bifilar magnet wire somewhere, with two contrasting insulation colors and a 50 ohm characteristic impedance. Traditional impedance value for twisted pairs of "wire wrap wire" is more like 110 ohms. It matter in baluns; I don't know about GDT's.
Today in a coilgun thread, Der Albi pointed out a cringeworthy video of magnet wire physical abuse.
Back to GDT's, has anyone tried bonded bifilar wire?
has a representative picture:I have a spool of untwisted bifilar magnet wire somewhere, with two contrasting insulation colors and a 50 ohm characteristic impedance. Traditional impedance value for twisted pairs of "wire wrap wire" is more like 110 ohms. It matter in baluns; I don't know about GDT's.
Wow! Those wires have some extreme insulation and that guy on the vid is !
but I can't put those on a 1" size GDT core :) maybe the thinner ones :) thanks for the comment. Re: GDT wire
kiat, Thu Oct 22 2015, 01:50PM
Triple insulated wire works well too, or small coax (eg. RG174 ) if you dont need too many turns
I wouldn't trust magnet wire
kiat, Thu Oct 22 2015, 01:50PM
Triple insulated wire works well too, or small coax (eg. RG174 ) if you dont need too many turns
I wouldn't trust magnet wire
Re: GDT wire
nzoomed, Thu Oct 22 2015, 09:09PM
I believe that UTP cable works quite well.
Im in the middle of constructing mine that will be used for steve wards universal driver.
Im unsure if im assembling it correctly though.
do you basically pair up each colour of wire?
I believe you parallel all the white wires together for the primary, is this correct?
nzoomed, Thu Oct 22 2015, 09:09PM
I believe that UTP cable works quite well.
Im in the middle of constructing mine that will be used for steve wards universal driver.
Im unsure if im assembling it correctly though.
do you basically pair up each colour of wire?
I believe you parallel all the white wires together for the primary, is this correct?
Re: GDT wire
Mads Barnkob, Fri Oct 23 2015, 05:14AM
That is correct, remember to mark the wires in one end of the cable before twisting the individual wires, so you know the phasing.
I used that method for all my GDTs. From driving TO-247 to CM600.
Mads Barnkob, Fri Oct 23 2015, 05:14AM
nzoomed wrote ...
I believe that UTP cable works quite well.
Im in the middle of constructing mine that will be used for steve wards universal driver.
Im unsure if im assembling it correctly though.
do you basically pair up each colour of wire?
I believe you parallel all the white wires together for the primary, is this correct?
I believe that UTP cable works quite well.
Im in the middle of constructing mine that will be used for steve wards universal driver.
Im unsure if im assembling it correctly though.
do you basically pair up each colour of wire?
I believe you parallel all the white wires together for the primary, is this correct?
That is correct, remember to mark the wires in one end of the cable before twisting the individual wires, so you know the phasing.
I used that method for all my GDTs. From driving TO-247 to CM600.
Re: GDT wire
Justin, Fri Oct 23 2015, 06:18AM
I like to tin the wires to mark them, but a sticker works fine too.
Justin, Fri Oct 23 2015, 06:18AM
Mads Barnkob wrote ...
That is correct, remember to mark the wires in one end of the cable before twisting the individual wires, so you know the phasing.
I used that method for all my GDTs. From driving TO-247 to CM600.
nzoomed wrote ...
I believe that UTP cable works quite well.
Im in the middle of constructing mine that will be used for steve wards universal driver.
Im unsure if im assembling it correctly though.
do you basically pair up each colour of wire?
I believe you parallel all the white wires together for the primary, is this correct?
I believe that UTP cable works quite well.
Im in the middle of constructing mine that will be used for steve wards universal driver.
Im unsure if im assembling it correctly though.
do you basically pair up each colour of wire?
I believe you parallel all the white wires together for the primary, is this correct?
That is correct, remember to mark the wires in one end of the cable before twisting the individual wires, so you know the phasing.
I used that method for all my GDTs. From driving TO-247 to CM600.
I like to tin the wires to mark them, but a sticker works fine too.
Re: GDT wire
nzoomed, Fri Oct 23 2015, 07:13AM
Well with cat5 its a little easier to mark phases, but im unsure what side is the "bottom" and "top" of the phase even if i do mark one side?
nzoomed, Fri Oct 23 2015, 07:13AM
Well with cat5 its a little easier to mark phases, but im unsure what side is the "bottom" and "top" of the phase even if i do mark one side?
Re: GDT wire
Goodchild, Fri Oct 23 2015, 11:27PM
You know I still use good old CAT5 on my GDTs. I have found over the years that nothing really beats it from a leakage inductance stand point. Those pre-twisted pars are great for GDT drive. Usually I will drive all the stripped windings in parallel while using the solids as my secondaries, this gets you a primary that's twisted with each individual secondary.
As other have said stay away from magnet wire with GDTs, remember that your GDT needs to able effectively isolate your bus voltage because when you turn on a high side and low side switch two of your winding will have the full bus voltage between them. Magnet wire is usually only good for a couple hundred volts of isolation and your bridge can be on the order of 350VDC or 650VDC in most cases. Magnet wire also doesn't like making 90* turn as it has to do around the core of a GDT. This can lead to breaking in your insulation and possibly crow-baring your gate with the full buss voltage.
Don't get to tied up in the shape and look of your waveform. What matters at the end of the day is how fast and effectively it turns on the gate.
Here are my suggestions:
1. For starters be sure to test your GDTs while actually connected to a gate. The type of load on the GDT will drastically effect the wave-shape. From your scope shots it doesn't look like your are driving a gate. Overshoot and rise time tend to improve with a load on the output.
2. I would select a different core, although N30 core has an initial permeability of 4300u it's saturating flux density is kinda low (0.38T) Check out T65 (0.46T) or some of the other flavors from TDK.
Goodchild, Fri Oct 23 2015, 11:27PM
You know I still use good old CAT5 on my GDTs. I have found over the years that nothing really beats it from a leakage inductance stand point. Those pre-twisted pars are great for GDT drive. Usually I will drive all the stripped windings in parallel while using the solids as my secondaries, this gets you a primary that's twisted with each individual secondary.
As other have said stay away from magnet wire with GDTs, remember that your GDT needs to able effectively isolate your bus voltage because when you turn on a high side and low side switch two of your winding will have the full bus voltage between them. Magnet wire is usually only good for a couple hundred volts of isolation and your bridge can be on the order of 350VDC or 650VDC in most cases. Magnet wire also doesn't like making 90* turn as it has to do around the core of a GDT. This can lead to breaking in your insulation and possibly crow-baring your gate with the full buss voltage.
Don't get to tied up in the shape and look of your waveform. What matters at the end of the day is how fast and effectively it turns on the gate.
Here are my suggestions:
1. For starters be sure to test your GDTs while actually connected to a gate. The type of load on the GDT will drastically effect the wave-shape. From your scope shots it doesn't look like your are driving a gate. Overshoot and rise time tend to improve with a load on the output.
2. I would select a different core, although N30 core has an initial permeability of 4300u it's saturating flux density is kinda low (0.38T) Check out T65 (0.46T) or some of the other flavors from TDK.
Re: GDT wire
Justin, Sat Oct 24 2015, 03:06PM
Surely twisted magnet wire has a lower leakage inductance, than cat5 cable.
Magnet wire can have a very high breakdown voltage depending on the grade.
I think the mechanical strain failure mode is an issue though.
Justin, Sat Oct 24 2015, 03:06PM
Surely twisted magnet wire has a lower leakage inductance, than cat5 cable.
Magnet wire can have a very high breakdown voltage depending on the grade.
I think the mechanical strain failure mode is an issue though.
Re: GDT wire
Goodchild, Sat Oct 24 2015, 03:52PM
The different in leakage inductance is marginal. Although you may be able to achieve a couple 10's of nH less leakage inductance with magnet wire it's not worth the large increase in probability of failure. This amount of leakage is really not going to show much improvement in the waveshape.
Goodchild, Sat Oct 24 2015, 03:52PM
Halfdead wrote ...
Surely twisted magnet wire has a lower leakage inductance, than cat5 cable.
Magnet wire can have a very high breakdown voltage depending on the grade.
I think the mechanical strain failure mode is an issue though.
Surely twisted magnet wire has a lower leakage inductance, than cat5 cable.
Magnet wire can have a very high breakdown voltage depending on the grade.
I think the mechanical strain failure mode is an issue though.
The different in leakage inductance is marginal. Although you may be able to achieve a couple 10's of nH less leakage inductance with magnet wire it's not worth the large increase in probability of failure. This amount of leakage is really not going to show much improvement in the waveshape.
Re: GDT wire
Hydron, Mon Oct 26 2015, 01:39PM
if you're unhappy with cat5 then use thin co-ax as mentioned above. Gives best possible coupling and probably more breakdown voltage capability.
Hydron, Mon Oct 26 2015, 01:39PM
if you're unhappy with cat5 then use thin co-ax as mentioned above. Gives best possible coupling and probably more breakdown voltage capability.
Print this page