TPS7H5001-SP: TPS7H5001-SP Flyback Reference Sheet Simulted in Simplis

Hi Mohamed,

I believe you may be correct regarding most of these changes. Regarding FZESR in row 83, the correct one to select depend on which of FZESR or RHPZ is lower. In the spreadsheet you attached above, FZESR is indeed lower so it can be used there instead of RHPZ, but they are very close to each other. I would recommend double checking your ESR input (G21) as lowering this from 10mΩ will lower FZESR and once again make RHPZ the lower of the two values.

As for your poor performance after making the edits, I believe that is because of an extra "G63" term left in the RCOMP calculation cell. When this term is removed the calculated RCOMP values should become much more reasonable. Since CCOMP  and CHF also depend on RCOMP this should correct those too.

One other thing I noticed that could definitely be messing with your simulation is the 11Ω current sense resistor. This will limit the current through the switch to only ~2.5A, and will therefore not allow the device to operate properly. You will certainly need a much smaller current sense resistor and also therefore a smaller current sense gain.

Thanks,

Andy

Hi Andy, 
Thanks for that! that definitely helped but not optimum yet.
with this correction we will have RCOMP 5,2K CCOMP 865nF CHF 3,39nF after simulating that we 50 Gcssp and 5 ohm RCS I got a really slow response with 90° PM & 40 dB GM, why is that ? is soemthing with calculation is wrong ?

when I did the same simulation with RCOMP 10K CCOMP 50nF CHF 300pF, I got 60° PM and 26 dB GM.




and how with 11 ohm we only allows 2.5A ?
RCS=VC_ILIM(1.05)/(IlPeak(10A)*Np(1)/Ns(100)=10.5 ohm 
decreasing the gain to 50 will result in RCS equal 5 ohm but the compensation network won't change, how comes the 11 ohm is affecting the converter operation ?

BR,
Mohamed

Hi Mohamed,

Your loop bandwidth actually looks very high based on the bode plot you attached... The target bandwidth should be between 1/4 to 1/10 of the right-half-plane zero frequency, which on the calculator file you attached previous was ~10kHz (so target crossover should be between 1 and 2.5kHz if that is still correct) "Switch-mode power converter compensation made easy" provides other good guidance also (Current-mode flyback is on page 27).

I think I just read your original post incorrectly, if you are using a 100:1 current-sense transformer then you should be okay. I was thinking of direct low-side current sensing with the current-sense resistor between the FET and GND, in that configuration the CS resistor can limit current through the switch.

Doing a quick simulation with the default SIMPLIS flyback schematic modified with a 100:1 CS transformer and a 10.5Ω CS resistor, the corrected excel calculator seems to provide reasonable values for a given FCO target. For example I selected 6kHz as a target and the simulation resulted in 5.3kHz FCO with good GM and PM. What is your target FCO if you are not achieving it? If you can attach your SIMPLIS file that may be helpful.

Thanks,

Andy

Hi Andy, 
my RHPZ is 10.8k, I choose 1k for crossover freq, it is in the sheet I sent to you before. 
one other thing is this peak at 249k, it looks similar to sub harmonic oscillations but it is happening at swicthing freq not Fsw/2 what can this be ? 
I noticed that when increasing RSC4-5 times the required value it disappear completely. 
I also notice that with my current compensation network the soft start time is not fulfilled.
I also notice that CS_ILim REACHES a really high voltage in the beginningTPS7H5001_SP_SIMPLIS_Flyback_Working_2Rails.zip how can this be dampened ? 

i can't choose FCO more than 2k it blinks red, how did you manage with my values to go for 6kHz FCO ?

file is attached.

Thanks! 
Mohamed

Hi Mohamed,

It looks like I was reading the color coding of your Bode plot backwards. Since green is Gain it does looks like Gain is coming in at ~1kHz, I am seeing the same when I run the simulation. The peak at 249kHz is caused by the switching frequency, you will see that peak on all switch-mode converters at the programmed switching frequency (if the Bode plot measures at the device switching frequency). As long as the crossover frequency is well below the switching frequency, it should not impact loop stability.

From what I can see by running a transient simulation of the SIMPLIS schematic you attached, it does look like the device is providing the configured soft-start. A 22nF cap is used on SS, which should configure ~5ms SS time, which aligns with what I am seeing in the simulation.

As for the spiking you are seeing, I think this is due to S2 being connected backwards. When I reverse the connection the spiking appears to go away:

Thanks,

Andy

Hi Andy, 
Thanks for the analysis. I did the change as you mentioned for the negative voltage in the Schematic. it solved the peak at CS_ILIM but Unfortunately this lead to a negative overall gain of the control loop. did you get a PM with this change ? 
I am a bit confused how the connection of the negative voltage should be, because with your connection the negative voltage winding turn number is the same as positive which is what to expect to get the same voltage and the CS_ILIM has a stable behaviour, on the other hand it leads to a negative overall gain of the control loop, not sure how is this related ?

one other question why is the cap ESR placed before not after the Cap to GND ? does it matter ?

I have the two networks one enabled and one disabled you probably got the 5ms SS with the original compensation network, not with the compensation network calculated in the sheet. I will attached again a file ready to be simulated with the compensation network values from the sheet exactly, you will find that the PM is not calculated and the converter takes more than 30ms to reach the expected output voltage. 
when I switch to R10,C10,C7 it's SS is  5.7ms 

is it expected that the calculation sheet values are not final ? how accurate is the simulation model ? should I rely more on the results in the simulation and take the calculation sheet just for the start ? can you list your recommendation

Thanks!
Mohamed

0753.TPS7H5001_SP_SIMPLIS_Flyback_Working_2Rails.zip

Hi Andy, 
I think you are right, Nearly all the problem are solved except that the system is overly compensated with PM 90°, I think I will need to do some tuning to see how can I get it to 72° PM.

Thanks! 
Mohamed

Welcome!

-Andy