Hi,
the internal regulator of the high torque models like SL1200ii, etc. provides for 21V to the circuitry.
For the consumer models with lower torque like SL5310 or SL2/3 its 12V.
Add 3V or more for the regulator itself and You are at 24V resp. 15V ... typical standard voltages for power supplies.
Higher voltages are neither required nor positive but just result in higher heat dissipation
I assume though that Technics designed with a way larger 'overhead' of more than 10V plus to allow for a larger ripple budget, hence the useage of a smaller smoothing cap and still achieve long lifetimes.
10W/5W supplies are fully sufficient, i.e about 0.5A.
I used small encapsulated SMPS from Meanwell, small enough to replace the original transformers.
No need to fuss around with a 'external' solution.
I run MCs and MMs and have no noise issues whatsoever ... and doubt very much that external overkill-design supplies do any better.
jauu
Calvin
the internal regulator of the high torque models like SL1200ii, etc. provides for 21V to the circuitry.
For the consumer models with lower torque like SL5310 or SL2/3 its 12V.
Add 3V or more for the regulator itself and You are at 24V resp. 15V ... typical standard voltages for power supplies.
Higher voltages are neither required nor positive but just result in higher heat dissipation
I assume though that Technics designed with a way larger 'overhead' of more than 10V plus to allow for a larger ripple budget, hence the useage of a smaller smoothing cap and still achieve long lifetimes.
10W/5W supplies are fully sufficient, i.e about 0.5A.
I used small encapsulated SMPS from Meanwell, small enough to replace the original transformers.
No need to fuss around with a 'external' solution.
I run MCs and MMs and have no noise issues whatsoever ... and doubt very much that external overkill-design supplies do any better.
jauu
Calvin
Well, if you check the datasheets of either the lm78xx or lm317 families, you will notice that these critters tend to like their Vin-Vout a bit larger. This is not directly addressed usually. But you can notice it in the graphs. For example in the latest TI LM317 datasheet, you will notice that ripple rejection vs Vout is shown/rated at 15V Vin-Vout. And it's 5dB higher than that shown in the ripple rejection vs current diagram that is at 5V Vin-Vout
I would hardly call a simple transformer with a CRCRC filter "overkill-design". If anything, the SMPS is probably more overkill, since not only it handles issues that are irrelevant in this DIY application (like efficiency or simplicity), but also ignores other more significant issues like noise that is literally transparent to the reg chips we are using.
As for how much the SMPS noise could couple audibly into the high gain stages of a phono, it would probably be hard to quantify with our limited DIY grade efforts. I could counter your observations with my own, but it would hardly be productive.
I want to point out though that a well designed linear psu is much more fit for the application and focuses on what matters. An SMPS focuses on things that are irrelevant. At the end of the day, I am sleeping better at night keeping SMPSs for applications where they make more sense
According to an earlier post, our SL1200s draw a max of 560mA upon start/stop.
Sometimes the outboard PS conversion is not an option, although it is theoretically best.
So we are left with internal PS and the options are:
1. SMPS with extra filters, such as AmyAlice . Would this also replace the internal regulator, setting the output to 21.6v dc?
Like this:
Traco Power TPP 15-124A-J
15w with adjustable output voltage. We can set the output to 21.6v and follow with AmyAlice filter.
The downside is that it does not have a "slow turn on" waveform like the original circuit.
But, it will not vibrate at 60hz or whatever your mains is.
2. We replace the original transformer with a toriodal of the same specs.
Like this:
Antek AN-0115 - 10VA 15V TRANSFORMER, can be wired for 30v.
Is 10va at 30v volts ok? Is better to use 24v? Original is 30v so it seems the best drop in with no other mods is 30v. Is this correct?
With the toroidal, we get no vibration at 60hz anymore, correct?
Feel free to discuss the 2 options above.... For the situations when external PS is NOT an option (due to various reasons)
Sometimes the outboard PS conversion is not an option, although it is theoretically best.
So we are left with internal PS and the options are:
1. SMPS with extra filters, such as AmyAlice . Would this also replace the internal regulator, setting the output to 21.6v dc?
Like this:
Traco Power TPP 15-124A-J
15w with adjustable output voltage. We can set the output to 21.6v and follow with AmyAlice filter.
The downside is that it does not have a "slow turn on" waveform like the original circuit.
But, it will not vibrate at 60hz or whatever your mains is.
2. We replace the original transformer with a toriodal of the same specs.
Like this:
Antek AN-0115 - 10VA 15V TRANSFORMER, can be wired for 30v.
Is 10va at 30v volts ok? Is better to use 24v? Original is 30v so it seems the best drop in with no other mods is 30v. Is this correct?
With the toroidal, we get no vibration at 60hz anymore, correct?
Feel free to discuss the 2 options above.... For the situations when external PS is NOT an option (due to various reasons)
I recently started modifying a second SL1200. This is a nice project to practice my Kicad, since Autodesk killed Eagle, and update my old PCBs while at it.
The schematic allows the reuse of common LM317 parts, just in case I like to add extra separate regs in the future for the motors or the rest of the electronics.
I have added a few optional helper parts like fuse/pptc, test points and separate led indicators for each stage of the reg. I have also added two 100nF caps, one before the LM317 since it's highly recommended when the cap bank is away from the chip, and a second next to the zener reference to further reduce its noise.
I have also added an output cap. The sl1200 PCB has one so I might remove or change this as I experiment.
The pcb is my old design that fits on top of the bottom entry of the chassis, and has an 8mm hole to fit a barrel power socket. The pads are intentionally oversized, which makes them easier to solder and test, while allowing for any resistor types like carbon composition or whatever.
I will delay ordering the PCBs while I refresh my memory on what we discussed here and at the art of sound forum, but I am pretty happy with the layout so far
Eventually I plan on making these available to my github
The schematic allows the reuse of common LM317 parts, just in case I like to add extra separate regs in the future for the motors or the rest of the electronics.
I have added a few optional helper parts like fuse/pptc, test points and separate led indicators for each stage of the reg. I have also added two 100nF caps, one before the LM317 since it's highly recommended when the cap bank is away from the chip, and a second next to the zener reference to further reduce its noise.
I have also added an output cap. The sl1200 PCB has one so I might remove or change this as I experiment.
The pcb is my old design that fits on top of the bottom entry of the chassis, and has an 8mm hole to fit a barrel power socket. The pads are intentionally oversized, which makes them easier to solder and test, while allowing for any resistor types like carbon composition or whatever.
I will delay ordering the PCBs while I refresh my memory on what we discussed here and at the art of sound forum, but I am pretty happy with the layout so far
Eventually I plan on making these available to my github
Rechecking my notes I did a few more changes.
I had an input cap in this board because I wanted to have the last RC stage close to the regulator, so that any effect from the cable's inductance is eliminated and that I have better transient response. The max diameter x height that probably fits under the platter is 35x30mm, which is enough to get us even 10mF@50V
I converted the small caps to 10mm so that I have more flexibility in sizes, and added a test point for the reference voltage.
I also added a spring clip for the TO220 reg. The pad on the chassis is thin and is getting pretty worn in my case from screwing/unscrewing. So I will just use this instead to push the reg to the metal. The PCB should be rigid enough.
Other than that, I added a few more 3d models to have a better overview of what the final board is going to look like. You can also see how the barrel jack looks like at the bottom, to have a better idea of what I am trying to do.
I am also thinking to add a MOV or a TVS to protect the rare chips from experimentation accidents
Any final notes before I send them to the fab?
I had an input cap in this board because I wanted to have the last RC stage close to the regulator, so that any effect from the cable's inductance is eliminated and that I have better transient response. The max diameter x height that probably fits under the platter is 35x30mm, which is enough to get us even 10mF@50V
I converted the small caps to 10mm so that I have more flexibility in sizes, and added a test point for the reference voltage.
I also added a spring clip for the TO220 reg. The pad on the chassis is thin and is getting pretty worn in my case from screwing/unscrewing. So I will just use this instead to push the reg to the metal. The PCB should be rigid enough.
Other than that, I added a few more 3d models to have a better overview of what the final board is going to look like. You can also see how the barrel jack looks like at the bottom, to have a better idea of what I am trying to do.
I am also thinking to add a MOV or a TVS to protect the rare chips from experimentation accidents
Any final notes before I send them to the fab?
Hi,
@#544 following:
Yes, the LM317 series is certainly not a low dropout design (LDO) ... anf it certainly is cheap and comes handy, but is not a good Reg in combination with SMPS, and You may consider more modern regulators that offer superior performance anyways.
The claimed rise in PSRR over Vin-Vout differential though is not shown in the TI´s most actual datasheet (SLVS044Y, revised 04-2020).
Figure7 shows a rising curve with generally rising voltage levels, but at a constant Vi-Vo dfferential of 15V.
Figure8 compares the effect of varying Cadj on PSRR over frequency at 5V Vi-Vo ... and there is obviously an error at least in the marking of those curves.
The blue curve should be the one with Cadj=10µF.
But still assuming otherwise identical conditions, at Vout=10V, there is hardly any difference in PSRR between those two diagrams, both about -65dB.
Also the level of interference depends on the distance between SMPS and circuitry.
Screening the SMPS with a metal shield (copper et al) should improve matters if neccessary.
Besides the same interference problem but right within the audible range applies to the low-frequency magnetic field of a transformer, but more difficult/costy to shield.
Technics at least put the transformer opposite of the bearing seen from the pick-up to achieve sufficient PS-to-PU distance.
Remains the wire-bound HF-noise/switcher residuals.
With SMPS You are almost noise-free within the audible range and just have to cope with the HF-hash.
I certainly wouldn´t use a LM317 as a post regulator for an SMPS .... as you said .... it´s simply too slow (PSRR drops at >10kHz) and all HF-denoising is almost put on the shoulders of the following Caps alone.
Since SMPS supply for quite constant output voltage anyway, we often don´t need additional voltage regulation, but either passive or active post-filtering is an adequate solution.
Passive LC-filtering and/or a Cap-multiplier will cope much better with the HF-noise .... maybe apart from those newer regulator designs of TI and ADI which keep their PSRR high up to 1MHz.
So, one could even use or modify the original Technics transistorized regulator as post-filter/cap-multiplier for an SMPS and still achive excellent results.
That focuses on what matters and keeps things simple.
A well designed linear PSU is certainly fit for the application ... but I´d like to point out its certainly no better than a well designed switching supply here ... but rather the opposite.
#546
just a few remarks to your layout:
The dynamic impedance and as such the noise should then still be lower than a voltage setting resistor/poti alone.
- the LM317 mounted horizontally at the PCb edge ..... the TO220 casing needs extra cooling in this application?
Is there a piece of sheet metal you could mount it to?
If not ... there is enough space on Your PCB to add a cooling fin and/or mount the LM317 vertical
(just noted that you changed the layout from #546 to #547)
jauu
Calvin
@#544 following:
Yes, the LM317 series is certainly not a low dropout design (LDO) ... anf it certainly is cheap and comes handy, but is not a good Reg in combination with SMPS, and You may consider more modern regulators that offer superior performance anyways.
The claimed rise in PSRR over Vin-Vout differential though is not shown in the TI´s most actual datasheet (SLVS044Y, revised 04-2020).
Figure7 shows a rising curve with generally rising voltage levels, but at a constant Vi-Vo dfferential of 15V.
Figure8 compares the effect of varying Cadj on PSRR over frequency at 5V Vi-Vo ... and there is obviously an error at least in the marking of those curves.
The blue curve should be the one with Cadj=10µF.
But still assuming otherwise identical conditions, at Vout=10V, there is hardly any difference in PSRR between those two diagrams, both about -65dB.
Well, a high gain phono stage could be susceptible to air-born HF-noise, but isn´t neccessarily prone to.
Also the level of interference depends on the distance between SMPS and circuitry.
Screening the SMPS with a metal shield (copper et al) should improve matters if neccessary.
Besides the same interference problem but right within the audible range applies to the low-frequency magnetic field of a transformer, but more difficult/costy to shield.
Technics at least put the transformer opposite of the bearing seen from the pick-up to achieve sufficient PS-to-PU distance.
Remains the wire-bound HF-noise/switcher residuals.
With SMPS You are almost noise-free within the audible range and just have to cope with the HF-hash.
I certainly wouldn´t use a LM317 as a post regulator for an SMPS .... as you said .... it´s simply too slow (PSRR drops at >10kHz) and all HF-denoising is almost put on the shoulders of the following Caps alone.
Since SMPS supply for quite constant output voltage anyway, we often don´t need additional voltage regulation, but either passive or active post-filtering is an adequate solution.
Passive LC-filtering and/or a Cap-multiplier will cope much better with the HF-noise .... maybe apart from those newer regulator designs of TI and ADI which keep their PSRR high up to 1MHz.
So, one could even use or modify the original Technics transistorized regulator as post-filter/cap-multiplier for an SMPS and still achive excellent results.
That focuses on what matters and keeps things simple.
A well designed linear PSU is certainly fit for the application ... but I´d like to point out its certainly no better than a well designed switching supply here ... but rather the opposite.
#546
just a few remarks to your layout:
- I´d place D2/R4 closer to the LM317 pins ... they should be in closest proximity to the LM´s pins.
- You could omit with C4. A low ESR of C3 should be good for PSRR at 100/120Hz, but if C4 is mounted C3 should rather be a high ESR-type.
- You might replace Zener-Diode Z1 by a Diode like 1N4148 to counter the temperature coefficient of Z2.
The dynamic impedance and as such the noise should then still be lower than a voltage setting resistor/poti alone.
- the LM317 mounted horizontally at the PCb edge ..... the TO220 casing needs extra cooling in this application?
Is there a piece of sheet metal you could mount it to?
If not ... there is enough space on Your PCB to add a cooling fin and/or mount the LM317 vertical
(just noted that you changed the layout from #546 to #547)
jauu
Calvin
You are correct. Thanks for pointing this out. I will try to find my old notes and get back to you
True. I'll try to juggle the parts a bit
Can you please elaborate? Will C4 demolish the low output impedance?
Yeah the temperature coefficient was mentioned in the acoustica articles. Plus with a second series part you can juggle them to trim the Vout to taste. Although to be honest the SL1200 is rather forgiving in its voltage requirements.
The pcb is designed so that the LM317 sits right on top of the old transistor's pad. I will post a photo of my current pcb later for clarity
Last edited:
I gave this a bit more thought, remembering an old article from TNT audio. While having R4 closer to the chip makes sense to allow it to develop the most stable voltage locally, this is not really our goal. Our goal is to have the most stable voltage at the load. Which means that ideally we would want the regulator to also take into account the PCB's layout and the wiring. The TNT audio article showed this only using "sense" wiring for the ground, but I wonder if it would make sense to use something similar for the V+ too. Larger bypass capacitors on the load could help further stabilizing things, but at the risk of reducing ESR too much and causing instabilities with the chip.
For reference
https://www.tnt-audio.com/clinica/regulators2_impedance1_e.html
To clarify how the PCB sits, here is my old one back from 2017. It started as a quick test to verify my measurements with a simple LM7820, but between work, losing interest and a few global crisis, it ended up spending almost a decade in my turntable.
Hi,
"Will C4 demolish the low output impedance?"
Paralleled the parasitic inductance and the capacitance form a LCR tank where R, resp. the ESR of the cap defines the Q.
Low ESR gives high Q, due to less damping, high ESR or even adding external R results in low Q.
In unfortunate high-Q cases this can show as a strong peak in impedance response, an antenna, susceptible to HF-noise.
Since the LM317 is so slow a good lytic alone will suffice.
Regarding 'sensing', I'd keep the protection diodes D1/D2 with both pins close to the LM317's pins and R4's Adj-side pin also close to the reg's Adj pin.
It's sensing side pin may then be routed with a thin trace or wire to the remote point.
jauu
Calvin
"Will C4 demolish the low output impedance?"
Paralleled the parasitic inductance and the capacitance form a LCR tank where R, resp. the ESR of the cap defines the Q.
Low ESR gives high Q, due to less damping, high ESR or even adding external R results in low Q.
In unfortunate high-Q cases this can show as a strong peak in impedance response, an antenna, susceptible to HF-noise.
Since the LM317 is so slow a good lytic alone will suffice.
Regarding 'sensing', I'd keep the protection diodes D1/D2 with both pins close to the LM317's pins and R4's Adj-side pin also close to the reg's Adj pin.
It's sensing side pin may then be routed with a thin trace or wire to the remote point.
jauu
Calvin
C4 is specified as 100nF though. Film or ceramic. Doesn't this mean that it lowers its ESR at a completely different frequency as the main electrolytic cap?
And also shunts any HF noise? (Including that of the zener)