Hi Salas
I'm using your Salas 1.1 design and the Pearl version 1 boards. The voltage reference and the shunt regulator is located on the Pearl board, the CCS and bias is hardwired off board. I was interested if it would be possible to "re-use" the Pearl board without spurious oscillations - seems to work but I will do additional tests.
Some more details: instead of the IRF9240 I'm using the IRF9640 - from the specs it appears to be equivalent except for the TO220. I couldn't find 9240 easily here. On the board I used a zener as reference. From the comments I've got the impression that at higher voltages this is a more stable solution. However from the first measurements I've got the impression that main part of the output noise is produced by the zener. Did anybody try to use a tl431 for example as a reference? Any comments are welcome.
Measurements were made at the reg's rail AC coupled under static load conditions using a resistor - about 36-37 mA. The amplifier part was not connected. Do you have any suggestions for test under dynamic load conditions? E.g. connect the amp and perform measurements while applying noise or while performing THD measurements?
brgrds,
Paul
I'm using your Salas 1.1 design and the Pearl version 1 boards. The voltage reference and the shunt regulator is located on the Pearl board, the CCS and bias is hardwired off board. I was interested if it would be possible to "re-use" the Pearl board without spurious oscillations - seems to work but I will do additional tests.
Some more details: instead of the IRF9240 I'm using the IRF9640 - from the specs it appears to be equivalent except for the TO220. I couldn't find 9240 easily here. On the board I used a zener as reference. From the comments I've got the impression that at higher voltages this is a more stable solution. However from the first measurements I've got the impression that main part of the output noise is produced by the zener. Did anybody try to use a tl431 for example as a reference? Any comments are welcome.
Measurements were made at the reg's rail AC coupled under static load conditions using a resistor - about 36-37 mA. The amplifier part was not connected. Do you have any suggestions for test under dynamic load conditions? E.g. connect the amp and perform measurements while applying noise or while performing THD measurements?
brgrds,
Paul
Paul:
The 9640 has good Ygfs but much worse Crss than 9240. 1.1 has a different calculation, with such parts you get a different OLG and Zo bandwidth, but less Zo. Also worse PSRR towards the highs if there is 9640 in the CCS too. Talking 9240 maybe you mean some other reg version? Yes the Zener can be more stable for drift, though with K117, a good resistor and just a small fixing trimmer, you may still find the drift acceptable, settles after ambient settles. The Zener needs 220R in series and a 220uF ecap across the lot to filter well, still some are much noisier than others and could need more capacitive filtering. Best are TC compensated, but not easy to get. TL431 is not great but better enough.
Your test is fine, just wanted to be sure the chart in Arta is set to logarithmic. View menu, frequency axis, log. Also 1/24 in the resolution just below. If the chart you showed was log, the trend shows the Zener which produces pink noise like noise spectrum. The noise floor limit of the audio card must be taken into account too. See with a simple loop through.
The 9640 has good Ygfs but much worse Crss than 9240. 1.1 has a different calculation, with such parts you get a different OLG and Zo bandwidth, but less Zo. Also worse PSRR towards the highs if there is 9640 in the CCS too. Talking 9240 maybe you mean some other reg version? Yes the Zener can be more stable for drift, though with K117, a good resistor and just a small fixing trimmer, you may still find the drift acceptable, settles after ambient settles. The Zener needs 220R in series and a 220uF ecap across the lot to filter well, still some are much noisier than others and could need more capacitive filtering. Best are TC compensated, but not easy to get. TL431 is not great but better enough.
Your test is fine, just wanted to be sure the chart in Arta is set to logarithmic. View menu, frequency axis, log. Also 1/24 in the resolution just below. If the chart you showed was log, the trend shows the Zener which produces pink noise like noise spectrum. The noise floor limit of the audio card must be taken into account too. See with a simple loop through.
Hi Salas,
Thanks you for the feedback! I'm a bit confused about:
Datasheets from IRF (IRFP9240 and IRF9640) mention a Crss of 81pF equal Rds(on) and a slightly different g. Comparing both datasheets I would think both devices are "equivalent" - or did I overlook something?
Regarding the zener, I'll setup an adjustable (5-10mA) CCS using 2sk170 and find out what happens to the noise level. Currently the zener is biased with a plain ccs and bridged with a 220uF and MMK - I didn't use a resistor in series with the zener. Next I'll try the adjustable resistor set-up.
brgrds,
Paul
Thanks you for the feedback! I'm a bit confused about:
Datasheets from IRF (IRFP9240 and IRF9640) mention a Crss of 81pF equal Rds(on) and a slightly different g. Comparing both datasheets I would think both devices are "equivalent" - or did I overlook something?
Regarding the zener, I'll setup an adjustable (5-10mA) CCS using 2sk170 and find out what happens to the noise level. Currently the zener is biased with a plain ccs and bridged with a 220uF and MMK - I didn't use a resistor in series with the zener. Next I'll try the adjustable resistor set-up.
brgrds,
Paul
You are right, I was remembering the IRF640 Crss. Still the SSLV1.1 uses different for Ciss. Anyway, since this is a bit of a spin off test proto, lets move to the simplistic regulators thread, the GB thread is not that adequate. Show me there the construction and any new Vref experiments. It does not oscillate shows good tolerance to different choices non the less.
.