Latest Developments at Belleson LLC
May 29, 2016—Highly Recommended for Digital
This post on DIY Audio talks about Belleson regulators used with DDDAC.
April 5, 2016—Less than -100dBV Noise and Ripple
Over the past 6 months we've been working on wideband noise and PSRR measurements. What we've learned is for something as good as our regulators, those measurements are really hard! Then on Monday I came across this application note from Linear Tech that describes how they've done it. Guess what? They say it's really hard!
On a different subject, our page describing the Dual Positive or Positive/Negative Supply has been updated to include a second pair of PCB files that are standard Gerber/Excellon data so you can order from multiple PCB vendors and not just ExpressPCB, They are known to work at http://www.oshpark.com/ and should be ok for other vendors. Also an additional transformer was added for higher output voltage (to 15V) with up to 1A current per supply.
February 29, 2016—Core Audio must be kidding!
It's leap day—let's leap into clarification of some bad information. Ryan Mintz of Core Audio has a special page to highlight Belleson Superpower regulators. While the part about Core Audio having used our products is true, the other parts are full of wrong and/or misleading information about the performance of our regulators. If you notice, there are no measurements, graphs or other hard facts, instead you find a lot of "arm waving" arguments about how their special proprietary design is somehow better at high frequencies than ours.
If that's true, why do people and manufacturers like them in their high end computer audio servers? Why are they used to power low jitter clocks? Why do we get customer feedback like this?
- So far, everyone who received my latest clock mod with SPM is amazed with the sound.
I've updated my system with the new clock module and I too am amazed with the improvements.
My [digital source] is now at a level in which I think few high-end DACs are able to come
close. It'd seem the SPM is now the perfect regulator for ultra low jitter clocks!
It's easy to talk about performance and hard to measure it. That's our specialty and how we know when our products are optimized. It's how we post so much technical information about how our regulators perform, and how our competitors perform. You want proof? Here are graphs of a noisy SMPS with 220kHz switch frequency and how it gets cleaned by a Superpower.
This is Vin to our regulator in time and spectrum:
and this is regulated Vout:
NOTICE that these graphs extend to 1MHz. Audio bandwidth? Yes. High frequency? Yes. And by the way, the internal loop amplifier we use (that Core Audio obviously does not know about) has almost 200MHz bandwidth.
We contacted Ryan mid-February and asked him to clarify and/or verify the information on the Core Audio page but have not received a reply.
And another thing—a regulator can't have dynamic load step response like a Belleson unless the circuit has extended frequency response. If you're at all familiar with Fourier series, you understand that the faster the edge, the higher are the frequency components it contains.
September 13, 2015—High Current Power Supply Design Note
This week we added a new FAQ entry to help select which of our 6 Superpower versions is best for your circuit, and a fabulous discussion of high current power supply design that examines almost every aspect of how to build a high current supply. It has a schematic, information on power dissipation, filter capacitor size, ripple, drop-out, a pretty waveform and a physical wiring diagram. It even has a protection circuit so you won't blow up anything if you accidentally short the output to ground!
July 4, 2015
This is a great tutorial on the function and application of series voltage regulators.
January 25, 2015 —Tempus Fugit!
Wow, six months got by in a hurry! We've been so busy doing things that there's been no time to blog about it. Back on May 1 2014 I mentioned a killer new super regulator product in the final stages of development. Those final stages turned into a lot of evaluation and testing, test and burn–in fixtures, new equipment, new PCBs, retesting, more evaluation, performance improvement, data sheet revisions, etc. etc. But that doesn't interest you, I'm just making excuses for all the time that passed.
What may interest you is the SPHP, a new 1000W Superpower that will power your computer music server with lots of pure, clean analog power. Another variation of the patented Belleson design that delivers big, with the same low noise, low output ripple, low output impedance and high speed step response as our other Superpower regulators.
It consists of a control PCB and a separate high power output transistor. The design allows high current to flow only through the power transistor, with only low driver current through the controller. This keeps load current induced transients away from the sensitive internal control circuitry to give the best possible performance.
Where can you get 10A at up to 100V? Right here!
This shows a Hafler DH-500 250W/ch power amp that's been upgraded with a positive and negative SPHP on each channel. The controller boards are mounted at the end of the cooling tunnel using aluminum channel attaching small PCBs. The power output transistors are mounted directly to the top of the tunnel. The regulators provide ±80V for both left and right channels at up to 3.5A per channel.
SPHP are available at output voltage from 5V to 100V, all able to drive 10+ amps with ease. See our latest datasheet for specifications and connection information
P.S. Christina saw this photo and said it looks like an automobile engine. Hah, that's right, put high performance parts in your audio engine!"
July 24, 2014 —Real World Results
Lorenzo Bucci modified the power supply of a Urai 1176 compressor/expander and used our regulators. His resulting measurements even surprised me!. He sent before and after noise floor FFTs (found on our Comments page) that show approximately 6dB improvement. Nice work Lorenzo!
July 11, 2014 —Another Pretender Exposed
A customer was curious about the performance of a switch-mode regulator from Murata, the 78xxSR, and was kind enough to send us a sample to test. We're always glad to get this kind of challenge because, in all honesty, the performance of Superpower amazes me every time.
This is a switching regulator that has an advantage of very high efficiency so uses very little power and does not need a heat sink. Like the ULN-HC78 below that was evaluated in April, we applied a 100mA step to see how well it regulates Vout.
A DC load of 22mA was applied and a 100mA step was applied on top of it. First, as soon as the 22mA DC load (540Ω Vout to ground) was applied, a 4mVpp 249kHz ripple appeared on Vout. The ripple amplitude increased to 10mVpp at 50Ω load. Adding the 100mA step looks like this:
Murata 7812SR with 100mA step applied
Top trace shows change in Vout as current switches from 22mA to 122mA (10mV per div). Note that ideally this trace should be flat! Bottom trace shows 100mA load current through 0.5Ω resistor (100mA per div)
Both photos were taken with the same test setup, same everything except the tested regulator. Efficiency is great if you're powering an ethernet interface that resends data with errors. But if you want clean fast Super power go to our order page.
July 6, 2014—A Superpower Supply Design
A power supply design is available on this new page. You can build your own dual power supply as either dual positive or positive+negative. The page has links to downloadable ExpressPCB files to order the PCBs, downloadable schematics for DipTrace and a parts list.
June 8, 2014—Updated Pages
May 1, 2014 — Another new Superpower product on the way!
Sneak preview—we have a killer new super regulator product in the final stages of development. I'll give you a hint—there may be a deity involved! Stay tuned.
April 10, 2014 —Superpower or Phantom Zone?
A customer told us about a new regulator available on eBay, ULN-HC78, so we bought one. It's touted as "Ultra Low Noise" and indeed it is. When we tested it, the noise was about 2µVrms and only 15mVpp, the lowest we've seen. It uses the lowest noise reference and op amp that Linear Tech makes, and has multiple capacitors to shunt any noise that may remain. So far, so good.
It's a regulator, riiight? We decided to see how it regulates Vout when a 100mA current step is requested from its output. Here is what we saw:
Top trace shows change in Vout as current switches from 0mA to 100mA (50mV per div)
Bottom trace shows 100mA load current through 0.1Ω resistor (200mA per div)
ULN HC78/12V with 100mA step applied
The applied step is 100mA for 40µsec, which you can see as the tiny blip on both traces at the center, left and right edges of the oscillogram. This small request for current causes tens of millivolts of ringing for more than twenty milliseconds! To put the time in perspective, 20msec equates to 50Hz, so the bouncing error generates non-harmonic content that covers almost the entire audio spectrum.
Both photos were taken with the same test setup, same everything except the tested regulator. What good is 2µV noise if it's covered by tens of mV of ringing? Which one looks Super and which one belongs in the Phantom Zone? After you decide, you can find the Super one on our order page.
December 1, 2013 — A New Low Current Lower Cost Superpower
Belleson is introducing the new SPM Superpower regulators.
October 12, 2013 — Updated Ripple Rejection Tests
We've updated the ripple rejection test data to be more accurate. The qualitative comparison data are similar to the first testing we did in 2011 but the results are more accurate by using a better calibrated measurement tool with input amplification to facilitate low level measurements.
September 23, 2013 — New test fixture for step response
Since 2010 we've said the droop in our dynamic output step response oscillograms was due to the test fixture. With the new higher performance Superpower we decided it's time to make a better fixture and prove it. We got a new test socket that clamps hard onto the DUT (device under test) pins, to eliminate socket related resistive voltage drops.
Then we got some 300MHz transistors for the dynamic active load on the DUT. We put them on a two–sided copper clad PCB very close to the new socket and attached some of those "Shakespearean Collar" heat sinks to the TO-39 cans. The fixture is ugly and performs very well.
Original Superpower (SPJ), 1A pulse
The original test fixture with a 1A current step. Bottom trace shows Vload across a ½Ω load resistor. Top trace shows change in Vout as current switches from 50mA to 1A. See the relatively slow vertical edges of the output trace and how the left edge of the step curves as it approaches its bottom? This was due to the slow load switching transistor. The new test circuit shows the real speed at which Superpower can deliver current.
New Superpower Regulator, 1A pulse
New test fixture with load = 1A step. The bottom trace shows Vload across a 0.1Ω load resistor. The faster switching transistor better shows the true speed of Superpower to deliver current and recover quickly. The top trace is the change in Vout. Notice the vertical scale here is 10 times closer than the original SPJ design photo on the left. Also the horizontal time scale is 2 times closer than the photo on the left and the edges are very fast.
Then we used a fast switching variable level square wave generator to switch the 300MHz transistors on and off. They are connected to the DUT via a 0.1Ω resistor that allows load current to be controlled as Vout/0.1. By varying the voltage to the base of the high speed load switching transistor, the load current can be set and viewed on an oscilloscope at 1A per division with vertical sensitivity of the 'scope set to 0.1V/division.
An example is shown here and you can see the difference in measured performance compared to the old fixture. Notice how the downward leading edge is vertical and the following horizontal portion of the waveform is very flat compared to the old fixture that used a 60MHz device.
September 1, 2013 — a new design with enhanced performance
The latest, greatest SPJ78 has new internal improvements to lower noise to an unprecedented level of <1ppm/°C. The device shown below has 12V output and measures 8µVrms over the audio bandwidth and about 60µV peak to peak. Combined with Superpower's stellar ripple rejection and dynamic performance, you can't find a voltage regulator with better overall performance.
Special Note: This measurement was made using the gain-of-1000 amplifier shown on our Noise page, with FFT from a Handyscope HS3. The vertical scale was adjusted by 60dB to account for the amplifier gain and the RMS and P-P units were changed from mV to µV. A Blackman-Harris window was used on the FFT and a 16 sweep average stabilizes the noise floor. Sample rate was 39.063KHz