Forums

Flyback primary inductance changes

Started by Piotr Wyderski August 4, 2018
I would like to build an off-line switcher based on the TinySwitch4 
family. It implies f=132kHz. Then, VOUT=15V@0.5A. V_IN_MIN=120V,
V_IN_MAX=374V (85-265VAC). I typed these values into the Schmidt's
calculator to get a rough estimate:

http://schmidt-walter-schaltnetzteile.de/smps_e/spw_smps_e.html

L_PRI=7.36mH. OK, so now use PI Expert Online, provided by the manufacturer:

Automatic design mode => TNY285DG, an EE13 core with L_PRI=1276uH,
0.25mm air gap. Almost 6 times lower than the previous value, okaaay...

Now, I don't like the proposed IC and replace it with a bit stronger
TNY287DG => L_PRI=497uH and a warning that the air gap is too long 
(0.745mm, yeah, right...).

The primary inductances differ just 15 times. What's going on here?

	Best regards, Piotr
On 08/04/2018 10:23 AM, Piotr Wyderski wrote:
> I would like to build an off-line switcher based on the TinySwitch4 family. It implies f=132kHz. Then, VOUT=15V@0.5A. V_IN_MIN=120V, > V_IN_MAX=374V (85-265VAC). I typed these values into the Schmidt's > calculator to get a rough estimate: > > http://schmidt-walter-schaltnetzteile.de/smps_e/spw_smps_e.html > > L_PRI=7.36mH. OK, so now use PI Expert Online, provided by the manufacturer: > > Automatic design mode => TNY285DG, an EE13 core with L_PRI=1276uH, > 0.25mm air gap. Almost 6 times lower than the previous value, okaaay... > > Now, I don't like the proposed IC and replace it with a bit stronger > TNY287DG => L_PRI=497uH and a warning that the air gap is too long (0.745mm, yeah, right...). > > The primary inductances differ just 15 times. What's going on here? > > Best regards, Piotr
inductance is probably chosen to get core loss and wiring loss to equal values. possibly the switch I^2R loss, aswell. The problem with those tools is that you never know what they calculate. Of course it might be complete garbage. In the end it is just one of those AI conspiracies to lure you onto the website to look at the ads.
"Piotr Wyderski" <peter.pan@neverland.mil> wrote in message 
news:pk3nr1$5up$1@node2.news.atman.pl...
>I would like to build an off-line switcher based on the TinySwitch4 family. >It implies f=132kHz. Then, VOUT=15V@0.5A. V_IN_MIN=120V, > V_IN_MAX=374V (85-265VAC). I typed these values into the Schmidt's > calculator to get a rough estimate: > > http://schmidt-walter-schaltnetzteile.de/smps_e/spw_smps_e.html > > L_PRI=7.36mH.
That sounds too high. Did you just take its default value? Did you click the Calc button and notice the waveforms? It recommends CCM. Default is based on 20% current ripple, I think. Nowhere near what a TinySwitch wants: DCM. Fortunately, you don't need to stick with the values given. You can plug in whatever value you like, and see the waveforms you'll get. :-) Roughly speaking, to get to 100% ripple (BCM), use 5x less inductance. Less even than that may be desired, because of the wide input voltage range. (Don't forget to check at the voltage extremes and see what happens!) Tim -- Seven Transistor Labs, LLC Electrical Engineering Consultation and Design Website: https://www.seventransistorlabs.com/
On 08/04/2018 04:58 PM, Johann Klammer wrote:
> On 08/04/2018 10:23 AM, Piotr Wyderski wrote: >> I would like to build an off-line switcher based on the TinySwitch4 family. It implies f=132kHz. Then, VOUT=15V@0.5A. V_IN_MIN=120V, >> V_IN_MAX=374V (85-265VAC). I typed these values into the Schmidt's >> calculator to get a rough estimate: >> >> http://schmidt-walter-schaltnetzteile.de/smps_e/spw_smps_e.html >> >> L_PRI=7.36mH. OK, so now use PI Expert Online, provided by the manufacturer: >> >> Automatic design mode => TNY285DG, an EE13 core with L_PRI=1276uH, >> 0.25mm air gap. Almost 6 times lower than the previous value, okaaay... >> >> Now, I don't like the proposed IC and replace it with a bit stronger >> TNY287DG => L_PRI=497uH and a warning that the air gap is too long (0.745mm, yeah, right...). >> >> The primary inductances differ just 15 times. What's going on here? >> >> Best regards, Piotr > inductance is probably chosen to get core loss and wiring loss to equal values. > possibly the switch I^2R loss, aswell. > The problem with those tools is that you never know what they calculate. > Of course it might be complete garbage. In the end it is just one > of those AI conspiracies to lure you onto the website to look at the ads. > >
Doing the math by hand isn't too hard, here is an instruction guide: <https://coefs.uncc.edu/mnoras/files/2013/03/Transformer-and-Inductor-Design-Handbook_Chapter_9.pdf>
"Johann Klammer" <klammerj@NOSPAM.a1.net> wrote in message 
news:pk541v$65j$1@gioia.aioe.org...
> inductance is probably chosen to get core loss and wiring loss to equal > values. > possibly the switch I^2R loss, aswell. > The problem with those tools is that you never know what they calculate. > Of course it might be complete garbage. In the end it is just one > of those AI conspiracies to lure you onto the website to look at the ads. > >
I can't speak for the PI tool, but the Schmidt tool has all the equations listed in the help box. It's just a calculator, nothing more: garbage in, garbage out. If your circuit does not conform to the assumptions made, if you don't get the waveforms seen; that's your problem. :) If you don't fill in the free variables yourself, it will recommend them, based on whatever assumption the writer chose to go with -- as it happens, CCM is assumed for all topologies I think. Tim -- Seven Transistor Labs, LLC Electrical Engineering Consultation and Design Website: https://www.seventransistorlabs.com/
On 08/05/2018 12:42 AM, Tim Williams wrote:
> "Piotr Wyderski" <peter.pan@neverland.mil> wrote in message > news:pk3nr1$5up$1@node2.news.atman.pl... >> I would like to build an off-line switcher based on the TinySwitch4 >> family. It implies f=132kHz. Then, VOUT=15V@0.5A. V_IN_MIN=120V, >> V_IN_MAX=374V (85-265VAC). I typed these values into the Schmidt's >> calculator to get a rough estimate: >> >> http://schmidt-walter-schaltnetzteile.de/smps_e/spw_smps_e.html >> >> L_PRI=7.36mH. > > That sounds too high.&#2013266080; Did you just take its default value?&#2013266080; Did you > click the Calc button and notice the waveforms?&#2013266080; It recommends CCM. > Default is based on 20% current ripple, I think. > > Nowhere near what a TinySwitch wants: DCM.&#2013266080; Fortunately, you don't need > to stick with the values given.&#2013266080; You can plug in whatever value you > like, and see the waveforms you'll get. :-) > > Roughly speaking, to get to 100% ripple (BCM), use 5x less inductance. > Less even than that may be desired, because of the wide input voltage > range. (Don't forget to check at the voltage extremes and see what > happens!) > > Tim >
Since it's a flyback it will have to be an air-gapped core of some type, whether distributed or "real", the saturation curve is more gentle compared to a ferrite brickwall. AFAIK there's no law that says you can't run a core a bit into saturation if it helps make the solution size smaller, for some e.g. powdered irons the curve looks like a complementary error function u can run four or five times the "cutoff" of 100% nominal relative mu vs H and only lose about 20% of the mu
On 08/04/2018 04:23 AM, Piotr Wyderski wrote:
> I would like to build an off-line switcher based on the TinySwitch4 > family. It implies f=132kHz. Then, VOUT=15V@0.5A. V_IN_MIN=120V, > V_IN_MAX=374V (85-265VAC). I typed these values into the Schmidt's > calculator to get a rough estimate: > > http://schmidt-walter-schaltnetzteile.de/smps_e/spw_smps_e.html > > L_PRI=7.36mH. OK, so now use PI Expert Online, provided by the > manufacturer: > > Automatic design mode => TNY285DG, an EE13 core with L_PRI=1276uH, > 0.25mm air gap. Almost 6 times lower than the previous value, okaaay... > > Now, I don't like the proposed IC and replace it with a bit stronger > TNY287DG => L_PRI=497uH and a warning that the air gap is too long > (0.745mm, yeah, right...). > > The primary inductances differ just 15 times. What's going on here? > > &nbsp;&nbsp;&nbsp;&nbsp;Best regards, Piotr
D_limit = 0.5 (current mode controller I believe) turns ratio = n_2/n_1 = (Vin_min * D_limit)/((V_out + V_diode) * (1 - D_limit)) = (120*0.5)/(15.2 * 0.5) ~= 8 D_ccm = ((V_out + V_diode) * 8) / ((Vin_min + (V_out + V_diode)) * 8) = ~ 10% Magentizing current referred to the primary = I_m = n_1/n_2 * 1/(1 - D_ccm) * I_out = 1/8 * 1/0.1 * 0.5 = 0.07A delta_Im = 0.2 * I_m = 0.014A L_m = (Vin_min * D_ccm * T_s)/(2 * delta_Im) = (120 * 0.1 * 7.57e-6)/(2 * 0.014) = 0.00324 Comes out to about 3.2 mH as an estimate for the required primary side inductance for CCM, how'd I do
On 08/05/2018 01:53 AM, bitrex wrote:
> On 08/04/2018 04:23 AM, Piotr Wyderski wrote: >> I would like to build an off-line switcher based on the TinySwitch4 >> family. It implies f=132kHz. Then, VOUT=15V@0.5A. V_IN_MIN=120V, >> V_IN_MAX=374V (85-265VAC). I typed these values into the Schmidt's >> calculator to get a rough estimate: >> >> http://schmidt-walter-schaltnetzteile.de/smps_e/spw_smps_e.html >> >> L_PRI=7.36mH. OK, so now use PI Expert Online, provided by the >> manufacturer: >> >> Automatic design mode => TNY285DG, an EE13 core with L_PRI=1276uH, >> 0.25mm air gap. Almost 6 times lower than the previous value, okaaay... >> >> Now, I don't like the proposed IC and replace it with a bit stronger >> TNY287DG => L_PRI=497uH and a warning that the air gap is too long >> (0.745mm, yeah, right...). >> >> The primary inductances differ just 15 times. What's going on here? >> >> &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;Best regards, Piotr > > D_limit = 0.5 (current mode controller I believe) > > turns ratio = n_2/n_1 = (Vin_min * D_limit)/((V_out + V_diode) * (1 - > D_limit)) = > (120*0.5)/(15.2 * 0.5) ~= 8 > > D_ccm = ((V_out + V_diode) * 8) / ((Vin_min + (V_out + V_diode)) * 8) = > ~ 10% > > Magentizing current referred to the primary = I_m = n_1/n_2 * 1/(1 - > D_ccm) * I_out = 1/8 * 1/0.1 * 0.5 = 0.07A
Should be 1/8 * 1/0.9 * 0.5 = 0.07A