GY6 Engine Dyno - Built, not bought! (w/ full DIY CNC computer control)

[Travis @ Buggy Depot]

You're looking at the dyno's progress as of July 2017

I started building this dyno in the summer of 2012, before I knew anything about hydraulics, electronics, or circuit board design. In the beginning, my motivation for building a dyno wasn't really connected to the business. I wanted to do it because I wanted to push the GY6 engine to its limits, find its breaking points, and make improvements based on real test data (not superstitions).

It's been a long project, and a very deep learning experience. If you'd asked me in 2012 how long it would take to build the dyno, I would have told you "2 months tops". That was somewhat true, as the physical frame and hydraulics were all functional at that point, but capturing good data out of the machine (with no computer) was too difficult. Now here we are in the 5th year. Dev time is extremely limited. On average, I've only been able to dedicate a couple days per month to developing the dyno. Life and building a company tends to get in the way of non-critical things.

As of July 2017, I have every detail worked out. All of the components on hand. All of the plans and diagrams laid out on the table. At this point the only issue is getting the time. Time to put it all together physically, electronically, and bring it to life by coding the software.

The 2012 "two month" version of the dyno (no computer)
At this point, I could simulate a load for up to 28hp. But I couldn't easily capture good data. This wasn't cutting it, as a dyno isn't a dyno without rock solid data. Just a fancy test stand.



Making it a true dyno with reliable data capture and automation control
Professional dyno suppliers wanted $3,000 for JUST the dedicated dyno computer alone, so developing one myself for my exact needs was the obvious decision.

Here is my custom Dyno Computer. On the left is my dyno computer fitted into an off-the-shelf panel box enclosure. On the right are the stepper motors, driver, and power supply. All of the computer-related hardware was less than $300, sourced from standard CNC online parts stores.



I found that the most advanced commercially available dyno computers (for small engines) are built with yesterday's technology and are lackluster on features unless you want to spend upwards of $40,000 on a turn-key machine! This may have changed in the last few years. I haven't checked.

Q: What will you do with the dyno when it's finished?
A: Test performance parts, publish the data here, do shootout tests, and piss off a lot of snake oil sellers!

Q: How much has it all cost, up to this point?
A: I'll tally up everything and edit this post later. A decent estimate would be $2,500 (stretched out over 5 years).

Q: Where do you source your parts?
A: Everywhere.

• The frame was an old $35 piece of junk I picked up at an estate auction.
• The engine cradle is a Yerf Dog GX150 swingarm.
• The mechanical parts were fabricated here in the BD machine shop on the mill, lathe and plasma table.

This project is entirely funded by your orders!
YOU make this possible. Thanks for your support!

Progress in chronological order:

• Features Overview
• The 2012 build - Frame fabrication, and the Hydraulic system
• Early lessons learned
• The Computer System (Wireless DAQ, Operator Control, and Automation setup)
• Stripping, painting, and rebuilding
• Hydraulic "working fluid" temperature control
• Automation and remote control (with stepper motors)

[Travis @ Buggy Depot]

Dyno Features (as of July 2017)

Non-techie Feature Overview:

• Made mostly of off-the-shelf parts.
• Up to 28 horsepower maximum.
• Fly-by-wire remote wireless control via laptop. Safety first!
• Now extremely reliable data capture and storage.
• WiFi connectivity. No wires!
• Can test any GY6 engine variant (125cc to 235cc).
• Can be easily moved and trailer mounted for dyno sessions anywhere.

(Slightly more) Technical Overview:

• Capable of a blazing fast 500 data sample captures per second!
• WiFi 802.11b/g/n WPA2 Encryption.
• Dedicated Linux-based capture server for extremely reliable data storage, logging, and other data intensive operations.
• Hydraulic load cell rated up to 3000psi.
• Fully automated hydraulic fluid temperature monitoring and regulation.
• Heavy Duty Oil-to-Water Heat Exchanger.

The full blown Linux data capture server may have been a bit overkill. But as a web programmer, it's one of the most familiar and "no research needed" parts of this build for me. The server approach opens many opportunities for future features, like really outstanding graphs and real time webcasts later.

Benchmark testing of the current design shows that we're able to capture over 500 samples per second. Earlier "no server" designs could only capture 50 samples per second, and had very limited storage. That would have worked, but again, setting up the server comes at very little actual cost.

More current pictures (July 2017)

[Travis @ Buggy Depot]

The 2012 build - Frame fabrication, and the Hydraulic system

- The frame was some type of ancient junk table I picked up for $35 at an estate auction
- Yerf Dog GX150 swingarm
- Hydraulic tank was made from an old water heater
- The rest is just bar stock, bearings, flanges, etc

Making the hydraulic pump flange.

I have no idea where this piece of scrap came from, but it worked. I used the rotary table and a 5/8" end mill. You'll probably recognize the spindle bolt in the center that I used as an arbor.









Looking back, I wish I had taken more pictures of the process of welding the swingarm to the frame.

The very first "working" version of the dyno. It was rough for sure, but it worked!

[Travis @ Buggy Depot]

Why a Hydraulic Brake dyno? Why not inertia?

Well in a nutshell, inertia dynos are good if you only want to test the power of an engine as a "snapshot". A brake dyno is able to test power output over long duration of time. And perhaps more importantly, a brake dyno can help test reliability and longevity as well.

After all, what good is making great peak horsepower, if your hopped up engine won't last more than a few rides? With the brake dyno, we can stress test engines for hours at a time instead of seconds.

I'll explain more, but keep in mind that this is all oversimplified for the sake of brevity.

Most times when we think of a dyno, an inertia type comes to mind. You roll your vehicle up onto a machine (or mount your engine) in such a way that spins and accelerates a weighted drum as quickly as possible. As soon as you hit redline, the test is over. This only lasts between 20 to 30 seconds. That's good for measuring power if you only plan to ride in bursts of acceleration on flat ground, but doesn't quite allow us to test engines the real way we ride all ride the 150's. There are all sorts of terrain scenarios to measure that an inertia dyno isn't capable of reproducing.

That's why a hydraulic brake dyno is great for testing the GY6 and developing upgrades. Since it acts as a very powerful brake, we can simulate all sorts of terrain and scenarios, even hill climbing. We can test the engine and transmission's responses to varying loads, and even hold a steady RPM under high load for long duration to test for specific problems. This is awesome for reliability and longevity testing especially for developing new and exciting big bore, stroker, high-compression, and/or turbocharged configurations.

How does it run?
On a dyno run with a hydraulic brake, the engine is locked in at WOT first under just enough load to keep the engine under red line. Then the hydraulic valve is closed progressively, to force the engine to lower and lower RPM while still held at 100% throttle if you want max HP/TQ readings. If you're testing for something other than maximum power, you have infinite control over the load and throttle being applied, so extremely fine tuning is possible any particular RPM, throttle position, and amount of load.

How is HP/TQ determined?
You (generally speaking) need three things on a hydraulic brake dyno.

1. RPM of the engine
2. RPM (and fluid displacement per RPM) of the pump
3. PSI of the hydraulic fluid.

The PSI gauge, RPM of the engine, and RPM of the pump are used to calculate the torque and horsepower produced by the engine at that moment.

Early lessons learned

1. Writing is slow!
Who would have thought? As it turns out, reading 3 separate gauges and writing them down to calculate later isn't a very effective data capture method.

2. Lack of resolution with a mechanical dial gauge
The resolution of mechanical dial gauges isn't a good fit for this application. 3,000PSI dial gauge are typically marked in 50PSI intervals. Just one tick means a big difference in HP/TQ numbers and can throw everything off. Standalone digital pressure gauges are very expensive and cost prohibitive.

The solution for us is a fully digital 0 - 5000psi Industrial Hydraulic Pressure Transducer sensor. The sensor has 2ms response time (1/50th of a second), and delivers ±0.25% full scale accuracy up to 5000PSI.



3. Without a good way to control Hydraulic fluid temperature, repeatability goes out the window
In my opinion, the true measure of any dyno is it's repeatability. Even with great sensors, the numbers can get thrown off on a hydraulic dyno by uncontrolled temperatures of the fluid. To correct this, I've added a oil-to-water heat exchange, and a microprocessor-controlled radiator system to reject heat and maintain hydraulic temperature withing a very narrow range even during extended runs.

4. Safety.
Even with shielding, I don't want to be facing an engine running 10,000RPM under maximum load. Some sort of remote control was necessary in my opinion.

[Travis @ Buggy Depot]

Building the dyno computer

---

Main Processor: Microchip 64Mhz PIC18F45K22

Responsible for Automation and Data Acquisition


Microchip PIC18F45K22 Quick specs:

• 64Mhz Processor (16MIPS)
• 44-pin (36 Input/Output Pins)
• 10-bit Analog-to-Digital Converter
• 5-bit DAC (unused, but reserved for later)
• 1024 bytes EEPROM for settings
• 64kB Instruction Space
• 3x External Interrupts
• 4x Interrupt-on-change

---

Co-processor: Particle WiFi Module (120Mhz ARM Cortex M3 processor)

Responsible for wireless communications between the main processor and the server:

Task 1. Live streaming data to the server
Task 2. Receiving and relaying operator commands from the server


Particle PØ Wi-Fi module Quick specs:

• STM32F205RGY6 120Mhz ARM Cortex M3 Processor
• Broadcom BCM43362 Wi-Fi chip
• 802.11b/g/n capable
• 1MB flash, 128KB RAM
• 18 Mixed-signal GPIO and advanced peripherals
• Open source design
• Real-time operating system (FreeRTOS)
• Soft AP setup
• FCC, CE and IC certified

---

Serial Peripheral Interface: Getting the processors talking over the SPI protocol

Getting the processors communicating with each other required a bit of hookup wire to the correct pins, and a voltage level translator in the middle. The processors communicate over a protocol called SPI.

SPI is a serial data protocol that allows the chips to share information at high speed, millions of bits per second. This is more than fast enough for the streaming sensor data and commands in real time.

Although I like the "DIY" look of the hookup wire, this will all be replaced with a custom PCB motherboard that I'm developing now. The motherboard will contain circuitry for all of the voltage translation circuits, as well as all Input/Output conditioning, filtering, and protection for the overall computer system.



SPI Signal Diagram
Here's a good diagram of what an SPI conversation "looks like" between processors. This the main processor transmits "S" and the co-processor transmits "F". This happens about 1 million times per second at speeds that I've tested so far.

---

The enclosure

---

Making a Wall Mount

Drawing up the wall mount


I'm happy with how it turned out!
The unit you see on the left is the hydraulic fluid heater control circuitry and solid state relays. I'll cover details about that side of the system in a later update.

[Travis @ Buggy Depot]

A question from my cross-post on BuggyMasters Forum:

This is some good stuff , Travis. you can definitely de-bunk a lot of myths. rumors and false claims though the use of a dyno. I have a overly simplified version of your dyno where the motor capabilities are measured through the motor's ability to produce hydraulic pressure at specified rpms. I can create a graph illustrating the motor output throughout the rpm range and horsepower is found through a math formula on a calculator. A crude but effective way to illustrate the effects of motor modifications.
You stated It's critical to tightly regulate the hydraulic temperature. I was curious as to why?

Keeping the hydraulic temperature regulated tightly helps to reduce variation in readings. With viscosity of the fluid changing with temperature, power readings at 150° will be inconsistent from readings at 175° (even if all other variables remain the same). So temperature swings can be a big problem when testing from one engine to the next. Or even on the same run if it lasts for more than a minute or two. I'm planning on some extended torture testing, so we need to be able to balance heat in and heat out.

But accuracy aside, probably more important (safety related) is how fast the temperatures shoot through the roof when under full load. It's scary. Back before I had any cooling at all, I won't say that I ever looked up during a run and had a tank full of smoking oil. Nope, never happened.

There are some other design constraints. The pump's maximum temperature rating is 180F, so I'm considering that our max limit for the system. Right now I'm shooting for an operating range between 150° to 160°, with 155 nominal. This will probably change as I get closer to the point of needing to make a final decision on which fluid (ISO 32 or ISO 22) I'm going to run.

My current working theory is that it is best to keep the viscosity as light as possible to reduce drag in the system (which I believe will rob power at higher flow rates without being measured). The pump's manual specifies an acceptable kinematic viscosity range in centistokes from 200cSt (very thick) to 6cSt. Water is 1cSt for comparison.

So I'm going for a viscosity range of 10cST to 15cST (some margin for safety), and will code the processor to compensate the HP calculation for the small viscosity differences across the 10 degrees of temperature range. Would like to get that down to 5 degrees, but won't know until firing it up and seeing how well the cooling system works. And adding a second radiator if needed.

Thankfully, viscosity changes are (more or less) linear with temperature. I traced over the graph below to show what I'm expecting for our temperature vs viscosity over a 10 degree range.

[Travis @ Buggy Depot]

Re-posting another question:

This is so cool....Would be awesome for tuning carbs and efi if you had an afr meter hooked up also.

Going with the PLX AFR wideband sensor kit. I picked up a Gen 4 kit back in 2012 when they were new. I believe they're on Gen 6 now.





I think that it's great that they designed it to be used with external data loggers. The PLX generates a 0 to 5v analog signal that corasponds to the A/F ratio in that moment. So reading A/F from my processor is a simple deal with little headache. I plan to use a diode array to protect my processor's input pin against over voltage from the PLX, but that's about it.

Here's the note in the manual on the 0-5v output for use with data acquisition. Well worth the money.



And here's the chart showing output voltage according to A/F, and the formula for the processor. This will be part of the real-time data recorded and sent to the operator display.



Handheld readout and stop box (Using the PLX touch display)

The PLX kit came with a nice touch gauge. It's not necessary to use, as my processor will be recording the raw data from the PLX sensor. However, the gauge would be useful on a handheld display hung on a pendant right next to the dyno. Along with an emergency stop button and an extra text display to show basic real-time information. This will be helpful so that I don't have to run to the desk when I'm troubleshooting something on the engine and need to glance at basic system info quickly.



I picked up a blank 6x4 enclosure from polycase and got to work:













Cutting an acrylic cover for the text display