The output impedance of your calibrator is? I'm not sure you get valid results with that method. You might want to try a power mosfet in series with a resistor, as a variable current source, and the mosfet driven by any sine wave generator, or by your scope calibrator. A mosfet with low Ron isn't hard to find. Fully open and you'll have the series resistor load draw a constant current. As the mosfet varies its R, it'll behave like a variable resistor.
The reduction in isolation has to be because of something else. Consider the ESR of the reference cap. This forms a divider network with the 10k resistor. Say it has an ESR of .1R. This makes a divider of 10k, which should give us an isolation of over 60db. So it can' be the filter network that's causing the reduction in isolation.
- keantoken
- keantoken
I hooked "them" up to my headamp... The sound is only slightly cleaner, and It doesn't seem less noisy, but it is much more enjoyable, and definitely less harsh. Unfortunately the new prototype of my headamp has stability problems, which were made worse by the regs. But without my signal generator I can't test the regs' stability. I guess fixing that is next on my list.
- keantoken
- keantoken
Well, I fixed my signal generator. I knew what the problem was. I've fixed it once before, when it needed a 50V 1500uF cap. I mistakenly use a 35V 2200uF cap. It worked for some time, then died. Looking at it again, I noticed another one of the 8 old 50V, 1500uF caps have died, this time in the counter, which is why it won't display frequency. I happened to have an old Nikon 4700u laying around, and it has the lowest ESR I have measured of any cap. Unfortunately I can't replace the other cap since I don't have a good enough cap, but this will do until the next cap dies.
- keantoken
- keantoken
I replaced the 100R resistor with a 2SK170BL, selected out of about 50 to have the highest Idss at 11.25mA.
It measured the same ripples and voltages (with my limited scope). It sounds a little different. The sound may have a bit more body, but I think yesterday the sound was mellower. I don't know yet if I prefer the jFET or not. I know my ears are not reliable after 24 hours, and there are many conditions affecting the sound. For example, the impedance may have been lowered, but I have 2 x 1.1uF film cap bypass 12cm away, so a lower impedance may react. I really don't know. More listening is required. I may replace the jFET with a 68R tomorrow and see if it changes to better or worse.
It measured the same ripples and voltages (with my limited scope). It sounds a little different. The sound may have a bit more body, but I think yesterday the sound was mellower. I don't know yet if I prefer the jFET or not. I know my ears are not reliable after 24 hours, and there are many conditions affecting the sound. For example, the impedance may have been lowered, but I have 2 x 1.1uF film cap bypass 12cm away, so a lower impedance may react. I really don't know. More listening is required. I may replace the jFET with a 68R tomorrow and see if it changes to better or worse.
Last edited:
I checked the regs with my newly fixed signal generator and everything seemed quite okay. They are even stable into films, so added local 560n decoupling to me headamp. Female voices are more controlled, more realistic.
Not to say that this the newest, bleeding edge regulator. To be fair, it has inferior specs compared to the shunt regulator that Salas and Ikoflexer have to offer. The design goals were low impact on proceeding circuitry (low dropout), and respectable output impedance. The output impedance so happens to be close to or below that of most electrolytics, which is a boon since an equivalent electrolytic would need to be very large to achieve the same bass response.
As it turns out, the specs are good enough to rival other offerings, but not so extreme that it can't be bypassed with high quality film capacitors. To me this implies I've hit a "sweep spot" in design.
Another boon is that since in the properly working circuit Vce of the devices can never reach above a few volts, it can work at virtually any input voltage regardless of transistor specs, which means low-power, fast BJT's can be used as long as the capacitors are rated appropriately.
- keantoken
Not to say that this the newest, bleeding edge regulator. To be fair, it has inferior specs compared to the shunt regulator that Salas and Ikoflexer have to offer. The design goals were low impact on proceeding circuitry (low dropout), and respectable output impedance. The output impedance so happens to be close to or below that of most electrolytics, which is a boon since an equivalent electrolytic would need to be very large to achieve the same bass response.
As it turns out, the specs are good enough to rival other offerings, but not so extreme that it can't be bypassed with high quality film capacitors. To me this implies I've hit a "sweep spot" in design.
Another boon is that since in the properly working circuit Vce of the devices can never reach above a few volts, it can work at virtually any input voltage regardless of transistor specs, which means low-power, fast BJT's can be used as long as the capacitors are rated appropriately.
- keantoken