Comparing both models, the ‘VAF’ (early voltage) of the Toshiba model (VAF = 6.8, 11.95) appears to be extremely low. The (Toshiba’s company?) person that created this model may have obtained the Early Voltage from the graph of the quasi-saturation region (low Vce region).
Why not try setting the Toshiba model's VAF to the same value as the Cabilio model and see if THD decreases?
But conversely, the Cabirio model TTC004B VAF=500 seems a little too high.
Valid points. We have a two builds with 2SC3503E / 2SA1381E with testing done confirming that the as-built units performed in line with the expectations from simulations.
We'd ultimately need to do the same exercise with TTC004B / TTA004B to confirm what simulations are suggesting. It's on my to-do list.
We'd ultimately need to do the same exercise with TTC004B / TTA004B to confirm what simulations are suggesting. It's on my to-do list.
There's a lot of information in the Build Guide. It occurs to me that this can get overwhelming. I'm working on adding a "Key Decisions" section at the start to distill down what you'll need to decide on to build this:
- PSU
- What V1 PSU voltage do you want to run and what size transformer do you need
- Which V2 PSU option will you use
- Component decisions
- What gain will you run
- What VAS Devices will you use
- Will you be driving difficult loads (TO-247 or TO-264 outputs)
- How to source parts not available from typical suppliers (JFETS and Lateral MOSFETs)
- Chassis
- How much bias do you want to run and it's impact on heatsink requirements
I have 3 pairs of the original boards remaining. Prices are in the first post. If you PM your address, I'll get a shipping quote.
Also, I'm working on a revised board with a few teaks to the schematic. But this is still a few weeks out (time to finish + time to produce the PCBs).
More details to come...
Also, I'm working on a revised board with a few teaks to the schematic. But this is still a few weeks out (time to finish + time to produce the PCBs).
More details to come...
Hi @brian92fs , thanks for all your work.
A few questions based on observations in LTSpice:
Thanks
A few questions based on observations in LTSpice:
- R7 runs at about 400mW at full power. Do you think 1W is sufficient? I think I would use at least a 2W resistor.
- R13 and R14 do not have to stand any significant power. Why did you choose 2W?
- C2 is shown as a polarized cap in the schematic, but up to +/-40mA current are flowing through it. I think a nonpolar would be better in that place?
Thanks
For an amplifier in my case it has a small output, but I would like to build it with the lowest gain and use it as a headphone amplifier for my low-ohm headphones. Has anyone tested it as a headphne amplifier? My headphone Denon AH-D7200 (25 Ohm).
Hi @mz543578854,
I'll check on R7. I remember assessing this and thinking 1W was fine. In typical use the amp is unlikely to be running at full power, so it's probably fine.
For R12 & R13 (you wrote 13 & 14 but I think you meant these), these are higher to account for observed dissipation under square wave testing. This is a bit of a safety approach. If the amp ever encountered high frequency oscillation, these would likely fail if rated lower. Is 2W necessary? Probably not in typical use cases. But 2W provides a bit of a safety margin. Also, if a builder wanted to test with square waves, they are absolutely necessary. If you do plan to test with square waves see section 7.4 for testing guidance.
The symbol for C2 is meant to represent a bipolar electrolytic. It lacks a "+" symbol. Looking at the schematic, I see that the other polar electrolytics on the schematic are missing the "+" symbol. This is a error. I'll get it corrected. Thank you for pointing this out.
For gerbers, I do not plan to publish them. I've put hundreds of hours into PCBs and would prefer to keep these private. I hope everyone understands.
I'll check on R7. I remember assessing this and thinking 1W was fine. In typical use the amp is unlikely to be running at full power, so it's probably fine.
For R12 & R13 (you wrote 13 & 14 but I think you meant these), these are higher to account for observed dissipation under square wave testing. This is a bit of a safety approach. If the amp ever encountered high frequency oscillation, these would likely fail if rated lower. Is 2W necessary? Probably not in typical use cases. But 2W provides a bit of a safety margin. Also, if a builder wanted to test with square waves, they are absolutely necessary. If you do plan to test with square waves see section 7.4 for testing guidance.
The symbol for C2 is meant to represent a bipolar electrolytic. It lacks a "+" symbol. Looking at the schematic, I see that the other polar electrolytics on the schematic are missing the "+" symbol. This is a error. I'll get it corrected. Thank you for pointing this out.
For gerbers, I do not plan to publish them. I've put hundreds of hours into PCBs and would prefer to keep these private. I hope everyone understands.
@HRDSTL,
For headphone use, please be aware that minimum 24V V1 PSU is required to use the V2 PSU. You could run lower V1 rails, but you'd need to omit the V2 PSU or run an external V2 PSU.
This is due to the input requirements for the DC-DC converters (U81 & U82). They need a 18 to 36V input. The 24V minimum V1 PSU allows the V1 rails to sag due to regulations issues and/or load and still be within the input range of the converters.
For headphone use, please be aware that minimum 24V V1 PSU is required to use the V2 PSU. You could run lower V1 rails, but you'd need to omit the V2 PSU or run an external V2 PSU.
This is due to the input requirements for the DC-DC converters (U81 & U82). They need a 18 to 36V input. The 24V minimum V1 PSU allows the V1 rails to sag due to regulations issues and/or load and still be within the input range of the converters.
Yes, typo. In LTSpice R12 runs with 1-2mW at full power, R13 only has some uW.
Wouldn't the MOSFETs likely die quickly as well in case of high frequency oscillations? But sure, does not harm (other than space) to use higher rated.
You could use lower rated components. Different designers have their own approach to these
Rod Elliott's P3A uses 0.5 watt for 35V rails
Bob Cordell uses 2W to 3W for 56V rails
Douglas Self uses 0.75 to 2.5W for 45V rails
The Wolverine uses 2W to 4W for 49V rails
If you wanted to use a lower value, 1W would be my recommendation. I considered 1W and 2W when spec'ing these but decided on 2W to be on the safe side.
Rod Elliott's P3A uses 0.5 watt for 35V rails
Bob Cordell uses 2W to 3W for 56V rails
Douglas Self uses 0.75 to 2.5W for 45V rails
The Wolverine uses 2W to 4W for 49V rails
If you wanted to use a lower value, 1W would be my recommendation. I considered 1W and 2W when spec'ing these but decided on 2W to be on the safe side.
These are the changes I am working on for Version 2 of the schematic and Rev 1 of the PCB:
For #7, refer to post 9 for information on different approaches to thermal protection. For my build, I plan on using an always-on 12V PSU to power some of the control circuits. One of these will be the thermal protection circuit which controls power to the amp boards. The PSU for this will be implemented as a Meanwell IRM SMPS. The ground for this must be returned to the SMPS ground, not the amplifier star ground (confirmed with bench testing). As a result, the grounding used in the original PCBs will not work. Rev 1 of the PCB addresses this.
- Combine Amp & V2 PSU onto one board. This simplifies construction, but still allows for the same alternative V2 PSU options
- Combining the boards allows the V2 PSU RC filter to be improved to a RCRC filter. This improves PSRR by 6dB at mains frequencies.
- LED indicators added for the main V1 rails.
- SOIC footprint removed for the JFETs. The TO-71 devices are preferred as they can use sockets allowing the devices to be flipped during the build to minimize DC offset resulting for imperfect matching in the monolithic JFETS.
- Add 75Ω R14 & R15 LSJ689 degeneration resistors. These split the 200Ω VR2 trimmer into two 75Ω resistors with VR2 lowered to 50Ω. This improves the sensitivity of DC offset adjustments (each turn of the trimmer is a smaller resistance change) and prevents VR2 from adjusting "too far". Large adjusts via VR2 should be avoided and will typically be an indicator that a problem exists somewhere else with the components used.
- C15 added for lead compensation (C15 across R7). This provides for an alternation compensation approach. Simulations show a small improvement if lead compensation is used rather than miller compensation.
- Connector J90 changed from 8 pin to 9 pin. The additional pin allows the ground from LM35DT (U90) to be returned separately.
- Silkscreen text size increased. The font size on the original boards was way too small for my aging eyes.
- Correction: C1 pads diameter increased from 0.9mm to 1.1mm.
- Correction: Silk screen errors for Q5/Q6 on the bottom silkscreen corrected.
For #7, refer to post 9 for information on different approaches to thermal protection. For my build, I plan on using an always-on 12V PSU to power some of the control circuits. One of these will be the thermal protection circuit which controls power to the amp boards. The PSU for this will be implemented as a Meanwell IRM SMPS. The ground for this must be returned to the SMPS ground, not the amplifier star ground (confirmed with bench testing). As a result, the grounding used in the original PCBs will not work. Rev 1 of the PCB addresses this.
Last edited:
I see some folks use 5W, but as this amplifier is only 30V I would go for 2W. This would be safe.
Regarding LSK489 and LSJ689.
I think maybe time for a Group Buy.
As can be seen in this page it is possible to buy directly from linearsystems:
https://www.linearsystems.com/distributors
I think maybe time for a Group Buy.
As can be seen in this page it is possible to buy directly from linearsystems:
https://www.linearsystems.com/distributors
Hi @lineup,
I was thinking about this the other day. I think the issue is the volumes required to get a reasonable discount.
However, kit's might be worth considering if it included the boards and harder to source parts like the JFETs and Lateral MOSFETs. Maybe the DC-DC converters too. Everything else is pretty typical and available from the regular suppliers. The main benefit here is probably a bit of savings from shipping.
I was thinking about this the other day. I think the issue is the volumes required to get a reasonable discount.
However, kit's might be worth considering if it included the boards and harder to source parts like the JFETs and Lateral MOSFETs. Maybe the DC-DC converters too. Everything else is pretty typical and available from the regular suppliers. The main benefit here is probably a bit of savings from shipping.
