Part 4 -- Gary is very meticulous -- if you've followed his writing. There's a trade-off with an E-Core or torroid. While, as he mentions, the E-Core is limited bandwidth and doesn't transmit line noise, it's also likely to have higher leakage inductance which may result in some diode ringing.
He also seems to be a "belt and suspenders" guy -- as he uses a bifilar choke on the power supply, which should knock down EMI
He also seems to be a "belt and suspenders" guy -- as he uses a bifilar choke on the power supply, which should knock down EMI
Yes. Fortunately this is easy to eradicate. All you need to do is measure the secondary leakage inductance*, plug it into Hagerman's formula, and install a snubber
- Rsnub = (1 / (2 * Zeta)) * sqrt(Lleak / Ctot)
Attached are some measurements made by diyAudio member DNi and posted to a thread in the Power Supplies forum of DIYA. They show an unsnubbed secondary winding, and then three different snubbers with three different damping factors (Zeta). The scope is triggered at the instant when the rectifier diode cuts off.
*This turns out to be rather easy (!).
_
Attachments
Last edited:
No, the load sensing was fully developed in the 95 article
Why don't you do some reading and form your own opinion?
Must faster and reliable than sifting through confusing and conflicting posts
Jan
I recommend you read Hagerman's snubber paper and also the Quasimodo snubber-design paper.
The most thorough exposition I've ever seen, is in Christophe Basso's textbook, excerpt below. However Basso is concerned with high frequency switching power supplies and the transformers used in HFSMPS. Plain ordinary 50Hz/60Hz mains transformers, probably do not require the same level of precision; they operate on a very leisurely timescale (milliseconds!) so there are no uncomfortable trade-offs between power and noise-reduction.
Basso is careful to isolate and extract ONLY the secondary leakage inductance; however it may be preferable to use the effective secondary leakage inductance in snubber designs, with the primary's leakage inductance reflected (thru the square of the turns ratio) into the secondary. A few SPICE simulations may help you decide which one you prefer.
The first way to skin the cat is (i) measure Lleak and Ctot; (ii) select a desired Zeta; (iii) calculate Rsnub using mathematics.
The second way to skin the cat is (a) connect the actual transformer, rectifier, and snubber together in a working circuit; (b) electrically stimulate the circuit and observe the oscillatory ringing on a scope; (c) adjust Rsnub potentiometer until ringing disappears. This is the "no-math" approach.
_
Attachments
Last edited:
I saw that, but I was not sure that it would still apply with new opamps, as in Jan's article, only the voltage reference mod was highlighted.
Anyway, I modified what in the current board are R2 from 10k to 5k and R5 from 4.99k to 2.5k (I actually soldered one more resistor of the same value on top of the existing one) but I still get the same results.
I think the AD817 should be more suitable than AD797 recommended in the article: is it correct ?
Thanks,
Davide
Anyway, I modified what in the current board are R2 from 10k to 5k and R5 from 4.99k to 2.5k (I actually soldered one more resistor of the same value on top of the existing one) but I still get the same results.
I think the AD817 should be more suitable than AD797 recommended in the article: is it correct ?
Thanks,
Davide
Last edited:
The ratio of R5/R3 sets the output voltage. R5 should remain 1K. Vout will be Vref*(1+ (R5/R3)). (I don't know why Jan put the parenthesis around R5 in the article.)
The 797 is not recomended for this application.
For the 5V version, check that the opmp you use is specified for 5V total (+/-2.5V) supply.
Jan