Antique Engines and Old Iron
[Home] - [HELP] - [Forums] - [Library] - [Photo Gallery] - [Groups] - [Classified Ads] - [Subscribe] - [Links] - [Books] - [Sponsors] -

Go Back   SmokStak > SmokStak® Vintage Electrical Equipment > Generators & Electric Motors General Discussion > Generac Generators (SEARS, etc.)
Forgot Password? Join Us!

Notices

Generac Generators (SEARS, etc.) Restoring, maintaining and operating Generac generators.

Generac Generators (SEARS, etc.)

Generac Voltage Regulator Reverse Engineered...


this thread has 51 replies and has been viewed 77425 times

Reply
 
Thread Tools Display Modes
  #31  
Old 02-08-2011, 09:15:39 PM
jimgrease jimgrease is offline
Registered-II
 
Join Date: Jun 2010
Location: Broomfield, Colorado
Posts: 35
Thanks: 3
Thanked 4 Times in 4 Posts
Default Re: Generac Voltage Regulator Reverse Engineered...

I see that Kevin replied while I was writing this post......if we only have to deal with +/-3 Hz then this may not be an issue. My sims are too course, I'll re-run at 63 Hz.

>>>>>>>
I ran some simulations on the feedback network today. First, let me say that running the generator fast enough to produce 70 or 80 Hz is unlikely. But this is just a computer simulation

Here are the numbers:
With the potentiometer set at ~10k (giving 10V at R3 with 170V sine wave peak with 60Hz input frequency)

frequency ___ R3 peak value at 0.9s simulation time and feedback signal constant (which doesn't happen in the actual circuit with the Op-Amp and UJT)
----------------------------------------------------------------------------------------------------------------------------------------------------
10 Hz ___ 22.3 V
20 Hz ___ 18.4 V
30 Hz ___ 15.4 V
40 Hz ___ 13.2 V
50 Hz ___ 11.4 V
60 Hz ___ 10.0 V *desired op point
70 Hz ___ 8.9 V
80 Hz ___ 8.0 V

Since the regulator actually tries to keep the feedback value close to 10V; here are the numbers if the engine was allowed to run at speeds other than 60Hz/3600 RPM:

frequency ___ corresponding generator output peak value at 0.9s sim time with R3 at 10V
----------------------------------------------------------------------------------------------------------------------------------------------------
10 Hz ___ 75 V
20 Hz ___ 95 V
30 Hz ___ 112 V
40 Hz ___ 131 V
50 Hz ___ 150 V
60 Hz ___ 170 V * desired op point
70 Hz ___ 190 V
80 Hz ___ 210 V

So this looks like a low pass filter. As the engine speed increases to 3600, the feedback network allows more of the signal to reach R3/C2. Going above 3600 (60 Hz) would cause an overvoltage condition.

I think C1 was chosen on purpose by Generac to provide this soft startup feature (as Kevin suggested). But I also think that C3 was chosen to slow the regulator response time so that engine over speeds don't result in significant over voltage conditions (when the Governor is busy reducing engine speed after loads are removed).

So the question now is: how big are the speed excursions when loads are removed from the generator output with typical and worst case loads? I don't have enough generator experience to hazard a guess on this question.

Here are some screen shots of the simulations (LTspice which is free from Linear Tech, diode models were from install library):
@ 60Hz:


@ 20Hz:


@ 80Hz:

Last edited by BTPost; 02-09-2011 at 03:21:30 AM. Reason: Fixed... low in place of high...
Reply With Quote
Sponsored Links
  #32  
Old 02-09-2011, 02:17:53 AM
jimgrease jimgrease is offline
Registered-II
 
Join Date: Jun 2010
Location: Broomfield, Colorado
Posts: 35
Thanks: 3
Thanked 4 Times in 4 Posts
Default Re: Generac Voltage Regulator Reverse Engineered...

Kevin described this filter correctly. As the frequency increases, less of the feedback signal makes it to R3. The regulator circuit will then attempt to increase the generator output to compensate.

Above the 60 Hz set point, this characteristic will result in an over voltage condition.

I'll toss out the idea that this could be corrected very simply by building a band pass filter on the input to the rectifier module (instead of the low pass).

Normally, a non-resonating band pass filter is not useful because they may have a serious attenuating effect (as an example: for a 1 Volt input perhaps only .5 volt is output in the desired pass band).

However, we need a large attenuator on this feedback signal as part of the design. So I'm wondering if a passive bandpass filter built with capacitors would have a flat enough frequency response at the peak to be useful from say 55-65 Hz?

Any additional attenuation from the passive band pass filter could be compensated for by adjusting the other resistor values.

I'll try simulating this, but if anyone has some additional thoughts, lets hear it

Last edited by BTPost; 02-09-2011 at 03:22:31 AM. Reason: corrected...
Reply With Quote
  #33  
Old 02-09-2011, 12:37:58 PM
Kevin K Kevin K is offline
Registered-III
 
Join Date: Jan 2005
Location: upstate New York USA
Posts: 3,860
Thanks: 3,519
Thanked 3,985 Times in 2,107 Posts
Default Re: Generac Voltage Regulator Reverse Engineered...

The bandpass filter idea is a good one, but in order to implement a high pass or band pass filter with passive components we need to use an inductor. At 60Hz, the size, cost, and difficulty obtaining the required values gets out of hand quickly.

The problem is if you implement the voltage roll off as the frequency goes down, the voltage rises as frequency goes up. We don't necessarily need a roll off on the high end, just a limit.

Here's another idea. Let's add another voltage sampling circuit without the roll off. Then add another comparator that will cut off drive to the SCR's when the voltage rises above a selected value, say 125V. This way you can have the roll off and a hard ceiling on the voltage. You can also use a faster attack time, as I did in the attached drawing, since this is just a limit.
Attached Thumbnails
Click image for larger version

Name:	HV Limit.jpeg
Views:	239
Size:	22.4 KB
ID:	99925  
Reply With Quote
  #34  
Old 02-09-2011, 10:54:36 PM
jimgrease jimgrease is offline
Registered-II
 
Join Date: Jun 2010
Location: Broomfield, Colorado
Posts: 35
Thanks: 3
Thanked 4 Times in 4 Posts
Default Re: Generac Voltage Regulator Reverse Engineered...

Kevin,

That is very nice, I like it. It has the advantage of being relatively robust to the tolerance of C1, and I think it might regulate the generator output with less voltage variations overall.

The disadvantage of course is the extra parts

I did spend some time today looking at changing the filter response of the attenuator/feedback network. The original low pass knee point is about at 5.5 Hz if I did my math right. I'm sure that knee point changes some with the +/-10% tolerance on C1, but since the regulator circuit operates well past that point on the -20db slope at 60 Hz, I don't think the tolerance of C1 affects the operation of the regulator (a good thing).

However, to do what I wanted, and change the filter response to flatten out around 55 Hz, the tolerance of the capacitors will probably come into play.

As you mentioned before, the filter response that is actually desired here would be one that drops at 20 db/decade and then flattens out around 55 Hz. The other idea that would work is actually a 'band-stop' filter that is flat enough around the 55-65 Hz frequencies.

Initially I was discouraged because I thought I would indeed need to build a LC resonant circuit (with huge unobtainable component values) to implement such a band stop filter response. However, after some searching on the net, I did have some success today.
Take a look at this band stop circuit:http://www.allaboutcircuits.com/vol_2/chpt_8/5.html

I simulated this starting with component values that matched our original C1 and related Resistor values. After a short time fiddling, I managed to produce this filter response (vertical cursor is at 60Hz):

The solid green plot is the response of the band stop filter after some tweaking. I didn't have to use any out of the ordinary cap values to achieve this. So, it might actually work

To visualize the original low pass filter response, imagine that the roll-off never stops, but just keeps on attenuating the signal as frequency increases.

But my band stop circuit is not ready for prime time just yet. We're just getting started with it, and this is not my area of expertise.

By the way, if you don't have a circuit simulator, but you have used spice in the past, I would highly recommend you download LTspiceIV from linear.com

I've found it very easy to use. You can draw your own symbols if necessary, but I've found all of the extra symbols I've ever needed on the yahoo group that covers this program.

Here are some tutorial links:
http://csserver.evansville.edu/~rich...als/LTspiceIV/
http://www.pkropik.com/storage/12022...tchercad_1.pdf
Reply With Quote
  #35  
Old 02-10-2011, 11:30:11 PM
jimgrease jimgrease is offline
Registered-II
 
Join Date: Jun 2010
Location: Broomfield, Colorado
Posts: 35
Thanks: 3
Thanked 4 Times in 4 Posts
Default Re: Generac Voltage Regulator Reverse Engineered...

I finished up a new version of the feedback network on the computer today. It works in simulation, but real world testing is necessary. There is a bunch I could say about this circuit, but let me just say I'm still weighing the trade-offs, and deciding what to do next.

Further research into the twin-T filter that I linked to in my last post revealed a modified version that uses less parts. Its called a bridged-T filter, and I think its more useful for our purposes here. Info on this circuit in these links:
http://www.drp.fmph.uniba.sk/ESM/twin.pdf
http://www.geofex.com/article_folders/eqs/paramet.htm

I simulated the text book version first, then I modified it to unbalance the response. I wanted to retain the low pass roll off up to about 50 Hz and then flatten out. I added one Resistor to accomplish this unbalancing of the response.

Then I worked the configuration into the regulator circuit feedback network. And finally I adjusted things to use standard value parts.

Here is a screen shot showing the original feedback network and the unbalanced bridge-T feedback network together. The filter response is shown as well with the 60 Hz point indicated by the vertical cursor. I hope you all can read this.


The big negative of this circuit is that the 10% tolerance of the capacitors might result in a shift in the location of the low point. It would not help to have that spot shift too far from where I've got it in this simulation.

If you see anyway to improve upon this, please post your ideas.

Here is some simulated output numbers to compare with my prior post:

frequency ___ corresponding generator output peak value at 0.9s sim time with R3 at 10V
10 Hz ___ 80 V
20 Hz ___ 125 V
30 Hz ___ 160 V
40 Hz ___ 170 V
50 Hz ___ 172 V
60 Hz ___ 171 V
70 Hz ___ 169 V

compare those with these from my simulations of the original circuit reposted below:

frequency ___ corresponding generator output peak value at 0.9s sim time with R3 at 10V
----------------------------------------------------------------------------------------------------------------------------------------------------
10 Hz ___ 75 V
20 Hz ___ 95 V
30 Hz ___ 112 V
40 Hz ___ 131 V
50 Hz ___ 150 V
60 Hz ___ 170 V * desired op point
70 Hz ___ 190 V
80 Hz ___ 210 V
Reply With Quote
  #36  
Old 02-13-2011, 08:19:30 PM
jimgrease jimgrease is offline
Registered-II
 
Join Date: Jun 2010
Location: Broomfield, Colorado
Posts: 35
Thanks: 3
Thanked 4 Times in 4 Posts
Default Re: Generac Voltage Regulator Reverse Engineered...

Weekend update here. This will be a longer post, so apologies in advance.

My parts order came in and I started breadboarding the actual circuit. I ordered 6 UJTs from a surplus supplier. These parts are supposed to be new old stock (NOS). First I built a relaxation oscillator to test the new UJTs. Of the six, only one worked reliably, which is strange since all 6 UJTs passed the basic resistance test. However the resistance tests are low current. When these parts were put in the oscillator circuit, the bad ones must have disconnected the emitter lead due to internal heat with real currents flowing in them.

I guess I should feel lucky I found one good one out of the bunch. I did look up the 2N4871 on the NTE cross reference list, and there is a substitute from NTE: NTE6410. It's $5 each from Jameco, which is also about how much I paid for one good UJT in my batch of NOS surplus.

Once I had the UJT portion working to my satisfaction, I went about building the Op-Amp circuit. I used my bench power supply set at 20V. With this setup I measured the following UJT output parameters whenever the feedback voltage is below the 10V reference from the voltage divider:

spike period: 950 us
spike peak amplitude: 8-9V depending upon load (I didn't have the SCRs connected so I adjusted R8s value)

Then I substituted a few 741 Op-Amps I had in my junk box. I was measuring the current consumption between the Op-Amps, here are the results:

original IC 20V total current draw: 9.3mA
with a Motorola MC1741C current: 9.3mA
with a RCA LM741CN current: 9.5mA
with a Signetics UA741CN current: 9.6mA

The Sig UA741CN Op-Amp performed the worst, besides consuming more current, the rail output of the Op-Amp was lower causing a longer period between UJT pulses at 1.06ms (1060 us).

This difference doesn't really matter, but it is interesting. The Motorola is the closest to the unmarked original. When choosing an Op-Amp for this application, I suggest avoiding the FET versions. The 741 is a BJT design that has good protection built in, and when we are dealing with 60 Hz frequencies, we don't need a fast part. If someone wants to suggest a more rugged Op-Amp for this application, I'd like to know about it.

With the 9.3mA current measurement, I could calculate a 2.1k resistance for the 20V load that the Zener must supply. So, I went back to my earlier transformer circuit to see if the lack of a filter circuit on the Zener regulator matters. I had to make some changes to use the 1N4747 and the 3.9k R5 (which I got in a 10 Watt version as that's all my local shop had on the shelf).

What I ended up doing is combing two transformers from my junk box to achieve voltages over 200V. Then I hooked up the bridge rectifier, followed by the 3.9k resistor, finally adding on the Zener with a 2.1k load resistor. Let me say that I'm glad I have the 10 watt 3.9k, it gets hot!

Just like before, this circuit still has the zeroing spikes whenever the rectifier output dipped back down to 0. Next, I added in series the field winding, and this time there was NO effect on the oscilloscope trace. I really couldn't see a difference between having the field in the circuit or not. Also, I did use a reverse biased diode to completely mimic the actual field connections.

I now think the lack of a filter cap on the Zener output is intentional. To investigate this configuration, I removed the 2.1k load and put the Op-Amp/UJT circuit in the semi-complete system. I used my bench supply to provide the 10V feedback signal that doesn't depend upon the Zener output.

This was worth doing, and the results were interesting. I think those of you who suggested that the lack of a filter cap was for timing/synchronizing purpose are right.

I turned on the persistence mode on my oscilloscope and recorded these traces for the condition when the regulator is just turning on with a single pulse out of the UJT oscillator.

Here is the case when there is no filter cap (second trace is the pulsing DC out of the rectifier showing the zero point):



Here is the timing on the first pulse when I used a 10uF filter cap on the Zener:


Clearly, the first SCR firing pulse is being timed to happen just before the zero crossing occurs. This timing is much tighter without a filter cap on the Zener regulator.

As the feedback voltage drops more and more, the UJT is allowed to fire earlier in the DPE sine period. Here is two more oscilloscope photos showing what happens as the feedback drops (note: the second trace in these photos is the Zener output):


Finally, UJT firing the SCRs most of the time due to low feedback signal:



Not sure what I'm going to do next, except ponder these results.
Reply With Quote
  #37  
Old 02-25-2011, 01:40:59 AM
jimgrease jimgrease is offline
Registered-II
 
Join Date: Jun 2010
Location: Broomfield, Colorado
Posts: 35
Thanks: 3
Thanked 4 Times in 4 Posts
Default Re: Generac Voltage Regulator Reverse Engineered...

Just a quick note here to say I hope there are still others working on this

I have been focusing on the following:

1) improving my LTspiceIV simulations of this circuit using the results from the last round of bench testing. It's going well and I think I'll be able to use this to ask: 'what if I do this?' questions.

2) I substituted a PUT for the UJT in my simulations since I couldn't find a 2N4871 model (I couldn't figure out how to modify the existing 2N2646 model either, they do not have the same trigger point). However, It would appear that the PUT is a more robust part, and it's still being produced. So I think it's a better choice for this application. Also, things might work better if the PUT isn't put in oscillation mode, but I'm still exploring that option.

3) I've been studying the Onan YD VR21 circuit design. It is indeed similar, but the differences appear to be important. I'm having some trouble understanding certain component values used in the Onan circuit. Perhaps my simulation work will allow me to explore this circuit further.

4) I think I found a circuit that will reduce the output during the unlikely event when the thermal switch opens. I'm thinking of using a LM311 comparator to perform this function and it's open collector output will be used to starve the UJT capacitor, preventing the UJT from triggering while the feedback voltage is below some set point.

I need to find a source for B&S head gaskets for this old 10 HP engine. Before putting this thing back together I want to check the condition of the cylinder and the valves.
Reply With Quote
  #38  
Old 02-25-2011, 01:00:07 PM
Kevin K Kevin K is offline
Registered-III
 
Join Date: Jan 2005
Location: upstate New York USA
Posts: 3,860
Thanks: 3,519
Thanked 3,985 Times in 2,107 Posts
Default Re: Generac Voltage Regulator Reverse Engineered...

Jim,

I'm still around, but I have not had much time to work on this project lately. I don't have a generator that uses a DPE winding, except my Onan BGD, and since that is one of my primary backups, I don't want to mess with it. The rest of the small generators use a "harmonic" winding with no voltage regulator, just a rectifier and capacitor connected to the armature. I've been looking on Craig's List for a cheap non working generator with a DPE winding and regulator, but so far no luck. I want to try and retrofit a VR one one of the generators with the harmonic winding, but another problem is the weather. Any generator I could test this on is out in a shed buried waist deep in snow, I just got another six inches this morning, and they are predicting another six inches on top of that. I think I'll wait for better weather so I can test something outside.

Checkout Tulas Engine Warehouse for the B&S parts.
http://tewarehouse.com/

If you need a service manual for the B&S flathead, I can email you a PDF copy.
Reply With Quote
The Following User Says Thank You to Kevin K For This Post:
  #39  
Old 02-26-2011, 01:13:34 AM
jimgrease jimgrease is offline
Registered-II
 
Join Date: Jun 2010
Location: Broomfield, Colorado
Posts: 35
Thanks: 3
Thanked 4 Times in 4 Posts
Default Re: Generac Voltage Regulator Reverse Engineered...

Kevin,

Yes, that would probably help me out to have a proper manual on hand. The engine on the generator is Model 251412. Date code starts with 78.

I looked at the TEW website but didn't get anywhere. I'll need a proper parts list for my engine to match up with the gaskets part numbers.

We've been lucky not to have that kind of snow around here this winter. I would wait for the melt too

Thanks.
Reply With Quote
  #40  
Old 02-26-2011, 09:33:01 AM
Kevin K Kevin K is offline
Registered-III
 
Join Date: Jan 2005
Location: upstate New York USA
Posts: 3,860
Thanks: 3,519
Thanked 3,985 Times in 2,107 Posts
Default Re: Generac Voltage Regulator Reverse Engineered...

The Briggs & Stratton website has IPL's - Illustrated Parts Lists - available. You will need your model, type, and code numbers.

http://www.briggsandstratton.com/eng...s-manuals.aspx

I just looked for your engine, and there are 12 different IPL's, depending on the code number.

I can email you a service manual if you PM me an email address that can accept large attachments. These used to be available on line, but apparently Briggs has pulled them from their own site, and requested they be taken down from others.
Reply With Quote
Reply


Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

F o r u m Jump

Similar Threads Chosen at Random
Thread Thread Starter F o r u m Replies Last Post
Generac 20kw Propane Regulator JBittnerSr Generac Generators (SEARS, etc.) 5 03-31-2011 01:32:28 PM
Voltage Regulator for MC-40 Generac golddredger Generac Generators (SEARS, etc.) 10 01-21-2011 12:50:10 AM
Generac 5.5 (RV) voltage surges jerseyjim Generac Generators (SEARS, etc.) 4 08-28-2009 01:29:17 AM
Generac voltage problems kphee Generac Generators (SEARS, etc.) 6 12-09-2008 12:57:51 AM
Generac regulator Andre Blanchard Generac Generators (SEARS, etc.) 2 02-06-2006 12:34:11 PM


Use "Ctrl" mouse wheel to change screen size.
All times are GMT -4. The time now is 03:58:03 AM.

Smokstak and Enginads site search!


All use is subject to our TERMS OF SERVICE
SMOKSTAK® is a Registered Trade Mark - A Community of Antique Engine Enthusiasts
Copyright © 2000 - 2016 by Harry Matthews P.O. Box 5612 - Sarasota, FL 34277