Prices for LDRs in this group buy will be $32.50. I normally charge 38.42 but in that 38.42 is included shipping and also a $3 donation to DIYaudio. This GB is doing the shipping for me and they are also donating to DIYaudio so my price drops by $6.
I have been talking to Andrew about the matching for 2SK170 and while I feel he has finally steered me in the right direction I dont want to offer you guys anything but the best.
LDRs are my realm and I do fantastic matching for them. 2SK170 might be easy but I want more experience matching them before I go offering them to you guys/gals.
His prices for the LSK170 are fantastic and the LSK170 I believe have better noise specs than the 2SK170.
Uriah
I have been talking to Andrew about the matching for 2SK170 and while I feel he has finally steered me in the right direction I dont want to offer you guys anything but the best.
LDRs are my realm and I do fantastic matching for them. 2SK170 might be easy but I want more experience matching them before I go offering them to you guys/gals.
His prices for the LSK170 are fantastic and the LSK170 I believe have better noise specs than the 2SK170.
Uriah
Note for those who fancy the Hypnotize as a symmetric shunt reg board mainly:
As I have posted enough times before, this symmetric shuntreg has tunings for the DCB1 and will perform its best as a pair. It will work as a general shunt very well, but bear in mind that it is optimized for DCB1. It has a purpose built Vref current assymetry. Specifically I exploit the njfet's assymetry to run the positive voltage reference harder than the negative, in order to match the step responses of the two sides. The positive Mosfets have different drive optimization and damping needs than the negative ones in conjunction with differences between their NPN/PNP drivers. You will invariably see a 2.5-3% +V prominence in average with matched leds. This is due to they run harder in the plus side. The slight Vout assymetry helps find better DC offset by swaping audio side njfets that are not terribly close by nature. Its economic, intuitive, does not take much knowledge or special rigs. There are positive subjective findings also about the whole evolution of the DCB1's pairing regs and my final choices, but I will leave that out, its wacko enough as a basis of technical conduct.
So, in a nutshell:
1. You use the regs for the DCB1 as it comes on the boards, fine. Forget about it, enjoy some tunes. Done the weird part 4U.
2. You want very close +/- symmetry for some reason? Only for other applications than the DCB1, up R5 on the positive side until they meet. Best will be half road. Use somehow less Vref total leds on the + side's 5 string Ref, and up its R5 a bit. It works like a faucet for the njfet under vbe and the mV across it.
3. Want more ICCS current for another application? Lower R1's. (3LedsVf-Vgs)/R1=ICCS. As it is, it should run 60-70mA. You may need +20% of the Ibias on class A circuit feed if conservative, but double or more is a bit better for ample dynamic reserve. It directly hits on dissipation, so sinks or Mosfet assembly underneath PCB a' la power amp style and bolding on metal floor or aluminum bar on insulation pads becomes mandatory.
4. If you want drastically different Vout for other applications, leave less leds in the Vref and add a resistor in the place of the missing one(s). Measure the drop across R5's prior to doing that, derive I, and calculate the resistor needed for target Vout by using Ohm's law. Its Vdrop will add on the remaining Leds. Don't use only a resistor, the less its value the better the damping ability of the Vsource. So leave enough part on the Leds.
5. If you don't dig the above, don't tweak, grab a beer and take it easy.
As I have posted enough times before, this symmetric shuntreg has tunings for the DCB1 and will perform its best as a pair. It will work as a general shunt very well, but bear in mind that it is optimized for DCB1. It has a purpose built Vref current assymetry. Specifically I exploit the njfet's assymetry to run the positive voltage reference harder than the negative, in order to match the step responses of the two sides. The positive Mosfets have different drive optimization and damping needs than the negative ones in conjunction with differences between their NPN/PNP drivers. You will invariably see a 2.5-3% +V prominence in average with matched leds. This is due to they run harder in the plus side. The slight Vout assymetry helps find better DC offset by swaping audio side njfets that are not terribly close by nature. Its economic, intuitive, does not take much knowledge or special rigs. There are positive subjective findings also about the whole evolution of the DCB1's pairing regs and my final choices, but I will leave that out, its wacko enough as a basis of technical conduct.
So, in a nutshell:
1. You use the regs for the DCB1 as it comes on the boards, fine. Forget about it, enjoy some tunes. Done the weird part 4U.
2. You want very close +/- symmetry for some reason? Only for other applications than the DCB1, up R5 on the positive side until they meet. Best will be half road. Use somehow less Vref total leds on the + side's 5 string Ref, and up its R5 a bit. It works like a faucet for the njfet under vbe and the mV across it.
3. Want more ICCS current for another application? Lower R1's. (3LedsVf-Vgs)/R1=ICCS. As it is, it should run 60-70mA. You may need +20% of the Ibias on class A circuit feed if conservative, but double or more is a bit better for ample dynamic reserve. It directly hits on dissipation, so sinks or Mosfet assembly underneath PCB a' la power amp style and bolding on metal floor or aluminum bar on insulation pads becomes mandatory.
4. If you want drastically different Vout for other applications, leave less leds in the Vref and add a resistor in the place of the missing one(s). Measure the drop across R5's prior to doing that, derive I, and calculate the resistor needed for target Vout by using Ohm's law. Its Vdrop will add on the remaining Leds. Don't use only a resistor, the less its value the better the damping ability of the Vsource. So leave enough part on the Leds.
5. If you don't dig the above, don't tweak, grab a beer and take it easy.
Hi Salas
I thought different voltages require Zener in place ( above +/-15V ) of LED as a reference ? I am ordering boards mainly to use Shunt with higher voltage and current and will ask you later on what exactly I can do to optimize but the resistor confuses my original idea of using low noise zener
Kannan
You can do both. If the zener is really low noise TC compensated, it has far lower impedance than using a resistor, the resistor guarantees very low noise, and the impedance can be just good enough, especially if the target voltage is half made by leds. Also zeners up to 4V7 or 6V2 the most, are certainly the lowest in noise of their grade, and you can use them in strings as leds, instead of a single 15Vz or higher. The 100uF bypass size in this reg and the DCB1 purpose pcb will not allow you space for a big bypass cap if the zeners you can find aren't exactly stellar. So, some leds and a resistor are the fastest and easiest but safer combination also, IMHO. There is a purpose designed pcb for reg, with connector for Vref freedom, connected as a subsystem (from leds to TL431), can be assembled as a plus or negative reg, has short thick star ground, etc.
1. Somewhere between 6 to 10 Idss.
2. Both R5 confirmed 10 ohms.
3. Neg rail 1.685, pos rail 1.770
I breadboarded a string of 5 LED's, 2sk170 and 10R resistor. On the working PCB, the voltages at the junction of the LED's and the 2sk170 is around 0.661V. The circuits for the pos and neg regulator are different in that the pos reg has the 2sk170 gate tied to drain, and the neg reg has the 2sk170 gate tied to source. So to get 0.661V at the junction of the 2sk170 and the LED's with gate tied to drain in the pos reg cct I adjusted the power supply voltage and ended up with approx 17mA through the LED's. So I guess 174mV across R5 in the pos reg cct *is* normal?
Salas, love the cct and the layout. I am not looking for ghosts, in the first instance I was trying to do some basic "health checks" to make sure all was within spec. So if I am chasing rabbits and using up forum bandwidth for no reason just let me know.
Maybe some basic voltage limits across strategic components suggested by you to confirm correct operation of the circuit will avoid many questions when everyone starts building? BTW the finished PCB really looks like a million bucks!
If you read again my explanation on the circuit choices, you will see that your DCB1 is absolutely OK. There I say I run the plus Vref harder and why. I explained how someone can get the regs closer if the offset is not a pig by upping the plus side R5. In general, if its not noisy, too hot, or having high DC offset, and the +reg is 3% more on average with matched leds on both banks and between them, plus side R5 has correctly more mV than neg side, is good to go. Your currents are OK judging by the Vdrops you gave me above.
P.S. Did you put it in audio service yet? The design has been done a long ago, and gives the effect that it gives. I did not feel it needed any more modification. Just try to get into it by using it in your stereo and see if you like it or not. Let it break in a little before paying serious attention, mainly the reg caps and main filter's come a bit more alive. You had a high mains trafo, did you help the ccs mosfets a little? Or they are just logically hot? Now that is something to cater for. Any new photos?
Just making sure that I am not confused: Vgs in your equation refers to Vgs of the K170 following the string [true/false].
On my Mezmerize board the Vgs for the one K170 that I measured is 3.7V. With a 5.7V LED string that amounts to just under 59mA.
On my Mezmerize board the Vgs for the one K170 that I measured is 3.7V. With a 5.7V LED string that amounts to just under 59mA.
Note for those who fancy the Hypnotize as a symmetric shunt reg board mainly:
3. Want more ICCS current for another application? Lower R1's. (3LedsVf-Vgs)/R1=ICCS. As it is, it should run 60-70mA. You may need +20% of the Ibias on class A circuit feed if conservative, but double or more is a bit better for ample dynamic reserve.
and take it easy.