This is low voltage headphone amp designed to drive 32R cans. The OPT is a multi-comp 120-0-120 0-12 0-12 1.6VA toroidal transformer to be cheap. The biggest problem is the primary capacitance with using mains transformers, which forms a dominant pole and limits the drive at high frequencies. Its for an FM tuner project so the HF is quite well controlled.
Comments please before I prototype would be appreciated (good or bad).
Comments please before I prototype would be appreciated (good or bad).
Attachments
OK just measured one - the simulation was a guess here. Primary total 500R so 250R per leg. Secondary 15R each so 7.5R in parallel - will update simulation.
Edit: a LTP push pull back end, ignore this post.
133mA so plenty, but it will be interesting to see the sim bode plot
133mA so plenty, but it will be interesting to see the sim bode plot
Last edited:
Operating CT push-pull but class A. The idea is that near equal current flow through the primary windings producing little magnetization. I understand that that's a better for a toroid otherwise there's a danger of core saturation. I think I add some 10R in series with the ECC88 cathodes to check how good the balance is, one side may have been made on a different day to the other.
Yep - the torroids for my V1 are setup to avoid saturation. Bigger cores for the VA.
I’m wondering if a cap attached version would also work. Like a Cuk smps.
I’m wondering if a cap attached version would also work. Like a Cuk smps.
Also how much cross talk are you going to get through the 12ax7 B+?
Worth duplicating and then running one channel 0V and the other with full input.
Worth duplicating and then running one channel 0V and the other with full input.
Thanks Nick for your comments. I've built a slightly modified version. Looks very good THD very low will measure tomorrow. Even though the mains transformers roll off at 8KHz due to the primary capacitance there's still 2Vppk available at 16KHz, and 30Hz, and the NFB corrects for the frequency response. It's FM so the stereo separation is at best 30dB with the LM4500A decoder. I am using the 12ax7 in the same way as a power amp so I guess the plate to plate capacitance is on the data sheet so easy to calculate and measure. The 3k3 and 22uF form a pole at .5Hz so I think we are OK with any coupling through the B+ into the 100k plate load - I was more worried about other things getting in on the B+ although it will be regulated.
I did see some coupling when overloaded due to the conduction on the grid of the ECC88 effecting the grid voltage and therefore the other channel - but only if overloaded. This could be improved by replacing R10 with a say 6v8 zener.
Very loud in 32R cans.
I did see some coupling when overloaded due to the conduction on the grid of the ECC88 effecting the grid voltage and therefore the other channel - but only if overloaded. This could be improved by replacing R10 with a say 6v8 zener.
Very loud in 32R cans.
Last edited:
Those transformers will pass 20khz single ended
But only 4khz push pull as the capacitance causes a complete phase change.
But only 4khz push pull as the capacitance causes a complete phase change.
I took the measurement with a sinewave generator + 10k across the full primary, this gave about -3dB at 8KHz with 32R secondary loaded. If you connect the sinewave generator to the transformer directly then you can get >20KHz bandwidth. This indicated a lot of primary capacitance presumably because the primary just goes round and round the core instead of being wound in sections.
It does indicate good coupling factor which is not what you get with a mains E type transformer - although I have not tried this. They are a fraction of the price (and size) of a proper headphone OPT.
It does indicate good coupling factor which is not what you get with a mains E type transformer - although I have not tried this. They are a fraction of the price (and size) of a proper headphone OPT.
Last edited:
Final version and prototype. THD is 0.3% at 2Vppk 1KHz which is very loud. Happy with this.
Attachments
Last edited:
Nice - looks like you subscribe to the RF school of pro typing!
Do you have a bode plot for the model vs what you have measured?
Do you have a bode plot for the model vs what you have measured?
Neat trick for cutri BFF islands: Circuit board 'Land' cutter. - YouTube by grinding a drill bit then using a bench press drill it allows you to make neat islands.